Abstract
Background
Cardiovascular-kidney-metabolic (CKM) syndrome represents a composite disease state driven by glucose and lipid metabolic dysregulation. The Cholesterol, High-density lipoprotein, and Glucose (CHG) index reflects the composite burden of these metabolic factors. However, its impact on the onset, stage-wise progression, and prognosis of CKM syndrome remains unclear.
Methods
This study utilized data from two large-scale prospective cohorts. In the UK Biobank (UKB) cohort, Fine-Gray proportional subdistribution hazards models were employed to investigate associations between CHG levels and the risk of incident cardiovascular disease (CVD), chronic kidney disease (CKD), and type 2 diabetes mellitus (T2DM), as well as the risk of progression from CKM Stage 0–1 to 2–3 and Stage 1–3 to 4. In the Beijing Anzhen Hospital cohort, Cox regression models assessed the impact of CHG on major adverse cardiovascular and cerebrovascular events (MACCE) in patients with CKM Stage 4 (established coronary artery disease). A causal forest algorithm was used to identify high-value subpopulations, and restricted cubic splines (RCS) characterized dose-response relationships. Sensitivity analyses were conducted to verify result robustness.
Results
The study included 370,916 participants free of CKM diseases from UKB (median follow-up: 16.5 years) and 8,494 patients with CKM Stage 4 from Beijing Anzhen Hospital (median follow-up: 645 days). In the general population, each 1-SD increase in the CHG index was significantly associated with an increased risk of incident T2DM (HR: 1.47; 95% CI: 1.40–1.53), CVD (HR: 1.07; 1.06–1.09), and CKD (HR: 1.07; 1.04–1.10). Furthermore, elevated CHG significantly accelerated CKM progression from Stage 0–1 to 2–3 (HR: 1.16; 1.11–1.23) and from Stage 1–3 to 4 (HR: 1.06; 1.05–1.08) per 1-SD increase. In patients with established CAD (Stage 4), a 1-SD increase in CHG was associated with a higher risk of MACCE (HR: 1.12; 1.05–1.19). Machine learning analysis revealed that this prognostic impact was particularly pronounced in patients with low systemic inflammation and elevated HbA1c.
Conclusions
The CHG index demonstrates significant predictive value for the onset of component diseases, stage-wise progression, and adverse prognosis across the entire CKM syndrome continuum.
Graphical abstract
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s13098-026-02192-2.
Keywords: Cardiovascular-kidney-metabolic syndrome; Cholesterol, high-density lipoprotein, and glucose index; Risk stratification; Coronary artery disease; Primary and secondary prevention
Introduction
Cardiovascular-kidney-metabolic (CKM) syndrome represents a complex clinical entity characterized by the progressive intersection and overlapping pathophysiological pathways of metabolic, renal, and cardiovascular diseases. In 2023, the American Heart Association (AHA) introduced the CKM framework, a staging system that organizes individual risk along a progressive continuum from Stage 0 (no risk factors) to Stage 4 (established cardiovascular disease) [1–3]. Lipid and glucose dysregulation trigger interconnected pathways, such as inflammation, oxidative stress, and endothelial dysfunction, that jointly promote kidney injury, atherosclerosis, and cardiac remodeling, thereby accelerating progression across CKM stages [4–8]. Epidemiological evidence shows that nearly 90% of U.S. adults meet criteria for CKM Stage 1 or higher, with 11.5% already progressing to Stage 4 [9–12]. Importantly, individuals in Stage 4 face significantly elevated risks of both cardiovascular and all-cause mortality compared to those in Stage 0, with hazard ratios (HR) of 3.88 and 2.03, respectively [10, 11, 13, 14].
While traditional markers—such as total cholesterol (TC), low density lipoprotein cholesterol (LDL-C), and fasting blood glucose (FBG), are commonly used for risk assessment, they capture only isolated aspects of metabolic dysfunction and often fail to capture the dynamic continuum of CKM syndrome, from early metabolic impairment to late-stage cardiovascular events. The novel Cholesterol, High-density lipoprotein, and Glucose (CHG) index, integrating TC, HDL-C, and FBG, was developed to capture the joint pathogenic impact of dyslipidemia and glycemic dysregulation [15, 16]. However, current studies on CHG have largely focused on isolated outcomes, such as T2DM diagnosis, stroke risk, hypertension association and kidney function decline [17–20]. It remains unexplored how the index is associated with the sequential progression of individuals through different CKM pathological stages. Furthermore, its prognostic value in patients with established CKM Stage 4 has not been fully evaluated.
Consequently, this study utilizes data from two large-scale prospective cohorts to systematically explore the life-course association between the CHG index and CKM syndrome. First, using the UK Biobank cohort (n > 400,000), we evaluated the predictive value of the CHG index for the initial onset of CKM syndrome and its progression across stages. Second, by analyzing a clinical cohort of 10,436 patients with established coronary artery disease (CAD) from Beijing Anzhen Hospital, we assessed the prognostic association between the CHG index and major adverse cardiovascular and cerebrovascular events (MACCE) in CKM Stage 4. Finally, a causal machine learning algorithm was employed to investigate effect heterogeneity across various clinical characteristics, aiming to identify specific subpopulations most likely to benefit [21].
Methods
Study design and eligibility criteria
We investigated the associations of the CHG index with the onset, progression, and prognosis of cardiovascular-kidney-metabolic (CKM) syndrome, utilizing data from two prospective cohorts.
To investigate the associations of the CHG index with the onset and progression of CKM syndrome, this study included participants from the baseline assessment (Instance 0) of the UK Biobank, a large-scale prospective cohort of over 500,000 individuals aged 40–69 years recruited across the UK (2006–2010) that received approval from the North West Multi-centre Research Ethics Committee (MREC), with electronic informed consent obtained from all participants. This research was conducted using the UK Biobank Resource under Application Number 103736.
To evaluate the prognostic value of the CHG index specifically in Stage 4 CKM syndrome, we analyzed a prospective cohort of patients with established coronary artery disease (CAD)—representative of this advanced stage—recruited from Beijing Anzhen Hospital. Between January 2021 and August 2024, we consecutively enrolled 10,436 patients with established CAD. Clinical management followed contemporary guidelines: all patients received guideline-directed medical therapy (GDMT) during hospitalization, and decisions regarding invasive strategies were based on standard indications [22]. Upon discharge, patients were prescribed standardized secondary prevention pharmacotherapy as part of their routine care. The study protocol was approved by the Institutional Review Board of Beijing Anzhen Hospital, Capital Medical University (Ethics Approval Code: KS2023081), and written informed consent was obtained from all participants.
Detailed inclusion and exclusion criteria for participants in both cohorts are presented in Fig. 1.
Fig. 1.
Flowchart of participant selection in the UK Biobank and Anzhen cohorts (A) UK Biobank and (B) Beijing Anzhen Hospital cohort
Data collection and definitions
Baseline characteristics, including sociodemographic factors, lifestyle behaviors, and cardiovascular-kidney-metabolic (CKM) profiles, were collected via touchscreen questionnaires, verbal interviews conducted by trained nurses, and physical examinations in UK Biobank. Baseline medication use (including lipid-lowering, glucose-lowering, and antihypertensive drugs) was ascertained through self-reported touchscreen questionnaires and nurse-led interviews (Fields 20003, 6153, and 6177). The staging of CKM syndrome was strictly aligned with the scientific statement from the American Heart Association (AHA) [1]. In accordance with the 2023 AHA Presidential Advisory framework, CKM Stage 3 was defined by a high predicted cardiovascular risk (10-year CVD risk ≥20%, calculated via the PREVENT equations). The AHA explicitly establishes this high-risk threshold as a clinical risk equivalent to subclinical cardiovascular disease, providing a guideline-endorsed operationalization for large cohorts where routine subclinical imaging is unavailable [23, 24]. The specific UK Biobank field IDs, along with detailed definitions and measurement protocols for all covariates, are listed in the Supplementary Tables 1–4.
Missing baseline covariates were addressed through multiple imputation by chained equations using the missRanger algorithm. This method utilizes an iterative Random Forest framework to predict missing values based on all other available variables. To preserve the natural distribution of the clinical data and avoid biologically implausible values, Predictive Mean Matching (PMM) was applied, where imputed values were selected from the nearest observed donors. The imputation process was configured with 1000 trees per forest and 5 iterations to ensure robust convergence.
Data Collection in the Beijing Anzhen Hospital Cohort In the Beijing Anzhen Hospital Cohort, data collection was conducted via standardized questionnaires administered by trained physicians blinded to the study aims. Baseline assessments comprised demographics, lifestyle factors, anthropometric measurements, and medical history (including comorbidities and prior revascularization). Hypertension and diabetes were ascertained based on self-reports, medication records, and clinical thresholds (e.g., fasting glucose and HbA1c). Routine laboratory analyses were performed at the hospital’s Clinical Laboratory Center following standardized protocols. Information regarding invasive procedures and post-discharge pharmacotherapy was extracted directly from electronic medical records.
The CHG index, the primary exposure of interest, was calculated identically in both cohorts using fasting blood parameters based on the formula [17]:
 |
Follow-up and outcome
We defined multiple endpoints to characterize the natural history of CKM syndrome. Cardiac (CAD, stroke, heart failure), renal (CKD), and metabolic (T2DM) diseases served as independent outcome events. Furthermore, to model disease progression, the transition from early CKM (Stages 0–1) to Stages 2–3 was defined by the incident diagnosis of T2DM, hypertension, or dyslipidemia. Finally, for the progression from Stages 1–3, the occurrence of any clinical CVD event marked the transition to Stage 4. Hospital admission data were obtained from Hospital Episode Statistics (HES) for England and Wales, and the Scottish Morbidity Record (SMR) for Scotland. Data linkage was performed by NHS England (formerly NHS Digital) and the NHS Central Register using participants’ unique National Health Service (NHS) identification numbers, dates of birth, and sex. Specifically, disease outcomes were defined based on primary or secondary diagnoses recorded in hospital inpatient records, coded according to the International Classification of Diseases, 9th and 10th Revisions (ICD-9 and ICD-10). Detailed definitions of outcomes are listed in Supplementary Tables 5–6.
Follow-up and Outcome Assessment in the Beijing Anzhen Hospital Cohort In the Beijing Anzhen Hospital cohort, prospective follow-up was conducted via telephone interviews scheduled every six months during the first two years post-discharge and annually thereafter. The primary outcome was a composite endpoint consisting of all-cause mortality, nonfatal myocardial infarction (MI), nonfatal stroke, and unplanned repeat revascularization. Secondary outcomes included: (1) a composite of all-cause mortality, nonfatal MI, and nonfatal stroke; (2) a composite of nonfatal MI and stroke; and (3) each individual component of the primary endpoint. Event occurrence and dates were initially ascertained through follow-up interviews and subsequently verified by reviewing medical records whenever available.
Statistical analyses
Descriptive statistics and baseline characteristics Baseline characteristics of the study population are presented as the mean ± standard deviation (SD) for continuous variables with a normal distribution, median (interquartile range [IQR]) for skewed continuous variables, and frequencies (percentages) for categorical variables. Participants were categorized into quartiles based on their baseline CHG levels. Differences in baseline characteristics across CHG quartiles were compared using analysis of variance (ANOVA) or the Kruskal-Wallis test for continuous variables, and the Chi-square test for categorical variables.
Survival analysis and association with outcomes Statistical approaches for survival analysis differed by data source. For the Anzhen Hospital cohort, cumulative incidence was visualized using Kaplan-Meier curves (compared via the log-rank test), and risks were estimated using Cox proportional hazards models. Conversely, for the UK Biobank cohort, we applied competing risk analyses to account for competing events. Cumulative incidence functions (CIF) were compared using Gray’s test, and Fine-Gray subdistribution hazard models were employed to determine risk estimates. Results are presented as hazard ratios (HRs) or subdistribution HRs with 95% CIs. Additionally, restricted cubic splines were utilized to assess potential non-linear relationships between continuous CHG levels and clinical outcomes. The discriminatory power of the predictive models was evaluated using the area under the receiver operating characteristic curve (AUC), with pairwise comparisons of the AUCs performed using DeLong’s non-parametric test. Furthermore, continuous net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were calculated to quantify the incremental predictive value and risk reclassification capacity of the new index beyond traditional baseline markers.
Sensitivity analyses Sensitivity analyses in the UK Biobank cohort included: (1) excluding events occurring within the first two years of follow-up; (2) using standard Cox proportional hazards regression instead of competing risk models; and (3) repeating the analysis on the original dataset without imputation (complete-case analysis).
Heterogeneity and causal machine learning To further investigate effect heterogeneity beyond standard subgroups, a causal forest algorithm based on generalized random forests was employed to estimate Conditional Average Treatment Effects (CATEs). This machine learning approach allowed us to identify specific patient characteristics that modify the association between the CHG index and MACCE risk. To mitigate overfitting, the model utilized an honest tree-building approach with optimized hyperparameters (1,000 trees, 4 variables per split, and a 0.7 subsampling fraction). To interpret the model and visualize the influence of individual features on the estimated CATEs, we utilized variable importance metrics and Partial Dependence Plots (PDPs). Variable importance quantified the relative contribution of each characteristic, while PDPs illustrated how the average predicted CATE varied with marginal changes in specific features, holding other covariates constant.
Statistical software All statistical analyses were performed using R software (version 4.5.0). A two-sided P-value of <0.05 was considered statistically significant.
Results
Baseline characteristics
The study included a total of 428,728 participants from the general population (UK Biobank) and 8,494 patients from a high-risk clinical setting (Anzhen Hospital). Baseline characteristics of the study population, stratified by CHG quartiles, are shown in Table 1. Overall, a graded worsening of clinical and metabolic profiles was observed with increasing CHG levels. Regarding sociodemographic factors, participants in the highest CHG quartile were predominantly male (SMD = 0.396) and exhibited lower socioeconomic status. Specifically, elevated CHG levels were associated with lower household income (SMD = 0.098), lower educational attainment (SMD = 0.103), and a higher likelihood of unemployment (SMD = 0.077). In terms of lifestyle, individuals with higher CHG levels were more likely to engage in adverse health behaviors, including unhealthy dietary habits (SMD = 0.077), physical inactivity (SMD = 0.122), poor sleep patterns (SMD = 0.056), and current smoking (SMD = 0.081). Metabolically, the high-CHG group displayed a significantly more adverse cardiovascular-kidney-metabolic (CKM) profile. These participants had higher body mass index (SMD = 0.510), elevated systolic blood pressure (SMD = 0.242), and poorer glycemic control (HbA1c; SMD = 0.336). Lipid profiles were also unfavorable, characterized by elevated total cholesterol (SMD = 0.506), LDL-cholesterol (SMD = 0.810), and markedly higher triglycerides (SMD = 0.950). Furthermore, higher CHG levels were linked to impaired renal function (lower eGFR; SMD = 0.203) and a higher predicted risk of cardiovascular events (PREVENT Score; SMD = 0.473).
Table 1.
Baseline characteristics of the general population stratified by CHG index quartiles
| n | level | Overall | Q1 | Q2 | Q3 | Q4 | p | SMD |
|---|---|---|---|---|---|---|---|---|
| 428728 | 107183 | 107181 | 107182 | 107182 | ||||
| Age, years, (mean (SD)) | 56.55 (8.09) | 55.42 (8.39) | 56.82 (8.04) | 57.12 (7.90) | 56.85 (7.91) | <0.001 | 0.105 | |
| Sex, n (%) | Female | 230742 (53.8) | 76133 (71.0) | 63560 (59.3) | 51622 (48.2) | 39427 (36.8) | <0.001 | 0.396 |
| Male | 197986 (46.2) | 31050 (29.0) | 43621 (40.7) | 55560 (51.8) | 67755 (63.2) | |||
| Ethnicity, n (%) | non-White | 18816 ( 4.4) | 4954 ( 4.7) | 4505 ( 4.3) | 4396 ( 4.2) | 4961 ( 4.7) | <0.001 | 0.017 |
| White | 404054 (95.6) | 100919 (95.3) | 101290 (95.7) | 101311 (95.8) | 100534 (95.3) | |||
| Socioeconomic factors | ||||||||
| Townsend Deprivation Index, (median [IQR]) | -2.15 [-3.65, 0.52] | -2.19 [-3.68, 0.42] | -2.21 [-3.69, 0.36] | -2.18 [-3.66, 0.46] | -1.99 [-3.56, 0.85] | <0.001 | 0.044 | |
| Household income, n (%) | Less than 18,000 | 83461 (22.8) | 18588 (20.2) | 20260 (22.2) | 21290 (23.3) | 23323 (25.5) | <0.001 | 0.098 |
| 18,000 to 30,999 | 92992 (25.4) | 21940 (23.9) | 23538 (25.7) | 23807 (26.1) | 23707 (25.9) | |||
| 31,000 to 51,999 | 95468 (26.1) | 24391 (26.6) | 23945 (26.2) | 23801 (26.0) | 23331 (25.5) | |||
| 52,000 to 100,000 | 74497 (20.3) | 20667 (22.5) | 18661 (20.4) | 18013 (19.7) | 17156 (18.7) | |||
| Greater than 100,000 | 19793 ( 5.4) | 6264 ( 6.8) | 5047 ( 5.5) | 4473 ( 4.9) | 4009 ( 4.4) | |||
| Marriage, n (%) | No | 37490 (10.8) | 10634 (12.2) | 9469 (10.9) | 8609 ( 9.8) | 8778 (10.1) | <0.001 | 0.042 |
| Yes | 310639 (89.2) | 76376 (87.8) | 77729 (89.1) | 78811 (90.2) | 77723 (89.9) | |||
| Employment status, n (%) | Employed | 247244 (58.3) | 65406 (61.6) | 61248 (57.7) | 60428 (57.0) | 60162 (56.8) | <0.001 | 0.077 |
| Unemployed | 34720 ( 8.2) | 8731 ( 8.2) | 7888 ( 7.4) | 8167 ( 7.7) | 9934 ( 9.4) | |||
| Retired | 142228 (33.5) | 31964 (30.1) | 36968 (34.8) | 37422 (35.3) | 35874 (33.9) | |||
| Education level, n (%) | High | 200927 (47.4) | 54835 (51.6) | 51307 (48.4) | 48603 (45.9) | 46182 (43.7) | <0.001 | 0.103 |
| Medium | 149280 (35.2) | 36356 (34.2) | 36813 (34.7) | 37777 (35.7) | 38334 (36.3) | |||
| Low | 73377 (17.3) | 14983 (14.1) | 17824 (16.8) | 19454 (18.4) | 21116 (20.0) | |||
| Lifestyle habits | ||||||||
| Alcohol intake, n (%) | Never | 18887 ( 5.3) | 3893 ( 4.2) | 4461 ( 5.0) | 4884 ( 5.5) | 5649 ( 6.6) | <0.001 | 0.098 |
| Previous | 15430 ( 4.3) | 2995 ( 3.3) | 3526 ( 3.9) | 3883 ( 4.4) | 5026 ( 5.8) | |||
| Safe | 168000 (47.2) | 45124 (49.2) | 43894 (48.9) | 41136 (46.4) | 37846 (44.0) | |||
| Dangerous | 153855 (43.2) | 39659 (43.3) | 37896 (42.2) | 38722 (43.7) | 37578 (43.6) | |||
| Dietary habits, n (%) | Unhealthy | 241771 (57.9) | 56965 (54.6) | 59300 (56.7) | 61443 (58.8) | 64063 (61.4) | <0.001 | 0.077 |
| Medium | 158863 (38.0) | 42579 (40.8) | 40914 (39.1) | 38923 (37.2) | 36447 (34.9) | |||
| Healthy | 17276 ( 4.1) | 4873 ( 4.7) | 4406 ( 4.2) | 4177 ( 4.0) | 3820 ( 3.7) | |||
| Sleep habits, n (%) | Unhealthy | 31221 ( 7.3) | 7040 ( 6.6) | 7332 ( 6.9) | 7753 ( 7.3) | 9096 ( 8.6) | <0.001 | 0.056 |
| Medium | 106563 (25.0) | 25400 (23.9) | 26242 (24.7) | 26952 (25.3) | 27969 (26.3) | |||
| Healthy | 287732 (67.6) | 74048 (69.5) | 72878 (68.5) | 71675 (67.4) | 69131 (65.1) | |||
| Physical activity, n (%) | Low | 61497 (18.5) | 12844 (15.2) | 14120 (17.0) | 15764 (19.1) | 18769 (23.0) | <0.001 | 0.122 |
| Moderate | 134566 (40.6) | 33675 (39.9) | 33737 (40.7) | 33926 (41.1) | 33228 (40.6) | |||
| High | 135461 (40.9) | 37774 (44.8) | 35085 (42.3) | 32827 (39.8) | 29775 (36.4) | |||
| Smoking status, n (%) | Never | 233158 (54.7) | 60698 (56.9) | 59763 (56.0) | 57858 (54.2) | 54839 (51.5) | <0.001 | 0.081 |
| Previous | 148158 (34.7) | 36354 (34.1) | 36796 (34.5) | 37428 (35.1) | 37580 (35.3) | |||
| Current | 45223 (10.6) | 9658 ( 9.1) | 10113 ( 9.5) | 11366 (10.7) | 14086 (13.2) | |||
| Cardio-Kidney-Metabolic indicators | ||||||||
| BMI, kg/m2, (mean (SD)) | 27.43 (4.79) | 25.26 (4.18) | 26.91 (4.50) | 28.09 (4.60) | 29.47 (4.81) | <0.001 | 0.51 | |
| HbA1c, %, (mean (SD)) | 5.46 (0.62) | 5.31 (0.41) | 5.37 (0.44) | 5.43 (0.45) | 5.72 (0.93) | <0.001 | 0.336 | |
| Glu, mmol/L, (mean (SD)) | 5.12 (1.24) | 4.65 (0.54) | 4.90 (0.54) | 5.06 (0.64) | 5.88 (2.08) | <0.001 | 0.57 | |
| SBP, mmHg, (mean (SD)) | 137.88 (18.64) | 133.39 (18.84) | 137.14 (18.56) | 139.47 (18.16) | 141.54 (17.99) | <0.001 | 0.242 | |
| TC, mmol/L, (mean (SD)) | 5.69 (1.14) | 5.18 (0.98) | 5.55 (1.03) | 5.85 (1.07) | 6.18 (1.23) | <0.001 | 0.506 | |
| LDL, mmol/L, (mean (SD)) | 3.56 (0.87) | 2.96 (0.64) | 3.43 (0.70) | 3.77 (0.77) | 4.07 (0.92) | <0.001 | 0.81 | |
| TG, mmol/L, (median [IQR]) | 1.48 [1.05, 2.15] | 0.99 [0.77, 1.30] | 1.30 [1.01, 1.73] | 1.67 [1.27, 2.22] | 2.33 [1.72, 3.17] | <0.001 | 0.95 | |
| eGFR, mL/min/1.73m2, (median [IQR]) | 96.08 [85.84, 105.52] | 99.57 [89.32, 108.22] | 96.50 [86.40, 105.77] | 94.69 [84.75, 104.07] | 93.61 [83.45, 103.24] | <0.001 | 0.203 | |
| Creatine, μmol/L, (mean (SD)) | 72.34 (18.44) | 69.04 (17.37) | 71.45 (18.10) | 73.52 (17.45) | 75.32 (20.10) | <0.001 | 0.19 | |
| PREVENT Score, (median [IQR]) | 0.06 [0.03, 0.10] | 0.04 [0.02, 0.07] | 0.05 [0.03, 0.09] | 0.07 [0.04, 0.10] | 0.08 [0.05, 0.12] | <0.001 | 0.473 | |
| Prescriptions | ||||||||
| Lipid-lowering drugs | 77755 (18.1) | 21718 (20.3) | 19942 (18.6) | 17454 (16.3) | 18641 (17.4) | <0.001 | 0.057 | |
| Glucose-lowering drugs | 16217 ( 3.8) | 2408 ( 2.2) | 2107 ( 2.0) | 2680 ( 2.5) | 9022 ( 8.4) | <0.001 | 0.151 | |
| Antihypertensive drugs | 100681 (23.5) | 22173 (20.7) | 24401 (22.8) | 25617 (23.9) | 28490 (26.6) | <0.001 | 0.074 | |
SD, standard deviation; IQR, interquartile range; SMD, standardized mean difference; BMI, body mass index; SBP, systolic blood pressure; HbA1c, glycated hemoglobin; Glu, glucose; TC, total cholesterol; LDL, low-density lipoprotein cholesterol; TG, triglycerides; eGFR, estimated glomerular filtration rate
Analyses in the Beijing Anzhen Hospital cohort corroborated the metabolic associations observed in the general population (Supplementary Table 7), particularly regarding the severity of dyslipidemia and glycemic dysregulation. In this high-risk clinical setting, the CHG index was strongly associated with impaired glucose homeostasis. Beyond biochemical profiles, elevated CHG levels paralleled disease severity and procedural complexity. Patients in the highest quartile more frequently underwent percutaneous coronary intervention (PCI) (66.0% vs. 58.2%; P < 0.001) and stent implantation (58.2% vs. 50.1%; P < 0.001). Furthermore, this group was managed with more intensive pharmacological regimens, characterized by higher usage rates of P2Y12 inhibitors (85.4% vs. 79.1%; P < 0.001) and beta-blockers (58.3% vs. 49.3%; P < 0.001).
Association of CHG index with incident cardiovascular-kidney-metabolic diseases in the general population
After excluding participants with prevalent CKM diseases at baseline, a total of 370,916 individuals were included in the final analysis. During a median follow-up of 16.5 years, we documented 14,271 incident cases of T2DM and 14,240 cases of CKD. A total of 39,893 incident CVD events were identified, which included 28,234 cases of CAD, 8,873 strokes, and 9,640 cases of HF. CIFs demonstrated a clear, graded stratification of risk across CHG quartiles for all cardiovascular-kidney-metabolic diseases. As illustrated in Fig. 2 (A–F) and Supplementary Table 8, participants in the highest CHG quartile (Q4) consistently exhibited the highest cumulative incidence of T2DM, CKD, and CVD, as well as specific CVD subtypes including CAD, stroke, and heart failure. The separation of curves occurred early during the follow-up period and widened over time, with statistically significant differences observed among groups (Fine-Gray test P < 0.001 for all outcomes). As shown in Supplementary Figures 3A-C, the CHG index demonstrated excellent calibration for predicting the 15-year incident risk of T2DM, CKD, and CVD, with the predicted probabilities being highly consistent with the actual observed event rates. Multivariable-adjusted analyses further confirmed the independent prognostic value of the CHG index (Fig. 2G and Table 2). After adjusting for confounding factors, elevated CHG levels remained significantly associated with an increased risk of all CKM outcomes. Comparing the highest versus the lowest quartile (Q4 vs. Q1), the association appeared most pronounced for T2DM [HR: 2.49, 95% CI: 2.31-2.70], followed by substantial risk increases for CVD [HR: 1.16, 95% CI: 1.12-1.21] and CKD [HR: 1.18, 95% CI: 1.11-1.27]. A series of sensitivity analyses yielded consistent results (Supplementary Table 9–11). Consistent results were obtained when modeling CHG as a continuous variable, where each 1-SD increase in the CHG index was associated with a significant elevation in the risk of incident events. RCS analyses elucidated the dose-response relationships between continuous CHG levels and incident outcomes (Fig. 2 H–J).
Fig. 2.
Association of the CHG index with the risks of T2DM, CVD, and CKD. (A–F) Cumulative risk curves for incident (A) T2DM, (B) CKD, (C) CVD, (D) CAD, (E) Stroke, and (F) HF across different levels of the CHG index. (G) Hazard ratios (HRs) and 95% confidence intervals (CIs) for the associations between CHG levels and the risks of T2DM, CKD, and CVD across four adjustment models. (H–J) Restricted cubic spline (RCS) curves illustrating the dose-response relationship between the continuous CHG index and the risk of (H) T2DM, (I) CKD, and (J) CVD. The curves are adjusted for covariates in Model 4. Notes: Models were adjusted as follows: Model 1 adjusted for age, sex, and ethnicity. Model 2 adjusted for covariates in Model 1 plus socioeconomic status (Townsend Deprivation Index, household income, marital status, and employment status). Model 3 adjusted for covariates in Model 2 plus lifestyle factors (alcohol consumption, dietary habits, sleep pattern, physical activity, and smoking status). Model 4 adjusted for covariates in Model 3 plus metabolic and clinical profiles (body mass index, systolic blood pressure, HbA1c, LDL-cholesterol, and triglycerides)
Table 2.
Association Between CHG and Cardiovascular-Kidney-Metabolic Diseases
| Model | Term | T2DM | CKD | CVD | |||
|---|---|---|---|---|---|---|---|
| HR (95% CI) | p | HR (95% CI) | p | HR (95% CI) | p | ||
| Unadjusted | Q1 | Ref | - | Ref | - | Ref | - |
| Q2 | 1.83 (1.71-1.96) | <0.001 | 1.25 (1.18-1.31) | <0.001 | 1.26 (1.22-1.30) | <0.001 | |
| Q3 | 2.94 (2.76-3.14) | <0.001 | 1.39 (1.33-1.46) | <0.001 | 1.58 (1.53-1.63) | <0.001 | |
| Q4 | 6.01 (5.67-6.38) | <0.001 | 1.63 (1.56-1.71) | <0.001 | 2.04 (1.98-2.10) | <0.001 | |
| per 1 SD | 2.03 (1.99-2.06) | <0.001 | 1.20 (1.18-1.22) | <0.001 | 1.31 (1.30-1.32) | <0.001 | |
| Model 1 | Q1 | Ref | - | Ref | - | Ref | - |
| Q2 | 1.73 (1.61-1.85) | <0.001 | 1.10 (1.05-1.16) | <0.001 | 1.07 (1.04-1.10) | <0.001 | |
| Q3 | 2.70 (2.53-2.88) | <0.001 | 1.20 (1.14-1.26) | <0.001 | 1.24 (1.20-1.27) | <0.001 | |
| Q4 | 5.51 (5.18-5.87) | <0.001 | 1.47 (1.40-1.54) | <0.001 | 1.55 (1.51-1.60) | <0.001 | |
| per 1 SD | 2.00 (1.97-2.04) | <0.001 | 1.16 (1.14-1.18) | <0.001 | 1.20 (1.19-1.21) | <0.001 | |
| Model 2 | Q1 | Ref | - | Ref | - | Ref | - |
| Q2 | 1.70 (1.59-1.82) | <0.001 | 1.08 (1.03-1.14) | 0.002 | 1.06 (1.03-1.10) | <0.001 | |
| Q3 | 2.62 (2.45-2.79) | <0.001 | 1.17 (1.11-1.23) | <0.001 | 1.22 (1.18-1.26) | <0.001 | |
| Q4 | 5.20 (4.89-5.53) | <0.001 | 1.40 (1.33-1.47) | <0.001 | 1.51 (1.46-1.55) | <0.001 | |
| per 1 SD | 1.95 (1.92-1.99) | <0.001 | 1.14 (1.12-1.16) | <0.001 | 1.18 (1.17-1.20) | <0.001 | |
| Model 3 | Q1 | Ref | - | Ref | - | Ref | - |
| Q2 | 1.69 (1.58-1.81) | <0.001 | 1.07 (1.02-1.13) | 0.006 | 1.06 (1.03-1.09) | <0.001 | |
| Q3 | 2.58 (2.42-2.75) | <0.001 | 1.15 (1.09-1.20) | <0.001 | 1.21 (1.18-1.25) | <0.001 | |
| Q4 | 5.04 (4.74-5.36) | <0.001 | 1.35 (1.29-1.42) | <0.001 | 1.48 (1.44-1.52) | <0.001 | |
| per 1 SD | 1.93 (1.89-1.96) | <0.001 | 1.12 (1.11-1.15) | <0.001 | 1.17 (1.16-1.19) | <0.001 | |
| Model 4 | Q1 | Ref | - | Ref | - | Ref | - |
| Q2 | 1.37 (1.28-1.47) | <0.001 | 1.03 (0.98-1.09) | 0.217 | 0.97 (0.94-1.00) | 0.080 | |
| Q3 | 1.77 (1.65-1.91) | <0.001 | 1.07 (1.01-1.13) | 0.026 | 1.04 (1.00-1.07) | 0.032 | |
| Q4 | 2.49 (2.31-2.70) | <0.001 | 1.18 (1.11-1.27) | <0.001 | 1.16 (1.12-1.21) | <0.001 | |
| per 1 SD | 1.47 (1.40-1.53) | <0.001 | 1.07 (1.04-1.10) | <0.001 | 1.07 (1.06-1.09) | <0.001 | |
Model 1: Adjusted for age, sex, and ethnicity. Model 2: Adjusted for covariates in Model 1 plus socioeconomic status (Townsend Deprivation Index, household income, marital status, and employment status). Model 3: Adjusted for covariates in Model 2 plus lifestyle factors (alcohol consumption, dietary habits, sleep pattern, physical activity, and smoking status). Model 4: Adjusted for covariates in Model 3 plus metabolic and clinical profiles (body mass index, systolic blood pressure, HbA1c, LDL-cholesterol, and triglycerides)
Association of CHG index with progression of cardiovascular-kidney-metabolic syndrome
Further analysis identified 119,245 participants at CKM stages 0–1 and 283,706 participants at stages 1–3. According to predefined outcome criteria, 3,208 individuals progressed to CKM stages 2–3, while 37,112 individuals advanced to CKM stage 4. As shown in Fig. 3 (A, B) and Supplementary Table 12, cumulative incidence functions revealed a significant, graded association between baseline CHG levels and disease progression. Participants in the highest CHG quartile (Q4) exhibited the highest risk of progressing from CKM stage 0–1 to stages 2–3 (Fig. 3A) and from CKM stage 1–3 to stage 4 (Fig. 3B). The divergence in risk trajectories among quartiles appeared early in the follow-up and persisted over time (Fine-Gray test P < 0.001 for both progression pathways). Supplementary Figures 3D-E indicate a strong agreement between the CHG-predicted risks and the true clinical outcomes. Multivariable regression analyses confirmed that the CHG index remained a robust and independent predictor of CKM progression after adjusting for sociodemographic, lifestyle, and clinical risk factors (Fig. 3, C and D and Table 3). In the fully adjusted model (Model 4), compared with the lowest quartile (Q1), participants in Q4 had a significantly elevated risk of advancing to more severe CKM stages. Specifically, the hazard ratio for progression from stage 0–1 to 2–3 was [HR: 1.68, 95% CI: 1.43-1.97], and for progression from stage 1–3 to stage 4 was [HR: 1.13, 95% CI: 1.08-1.18]. A series of sensitivity analyses yielded consistent results (Supplementary Table 13–15). Similarly, when modeled as a continuous variable, each 1-SD increase in the CHG index was independently associated with disease progression in both scenarios. Restricted cubic spline analyses characterized the dose-response relationship between the CHG index and CKM progression (Fig. 3, E and F). According to Supplementary Table 17, the CHG index demonstrated significant incremental value in predicting CKM stage progression compared with the individual application of TC, HDL-C, or GLU. Notably, regarding the progression from Stage 1-3 to Stage 4, although the CHG index did not yield a significant increase in AUC compared to HDL-C alone, the significant NRI and IDI results indicated a substantial improvement in risk reclassification and discriminatory power.
rFig. 3.
Association of the CHG index with the progression of CKM syndrome stages. (A, B) Cumulative incidence curves for the progression of CKM stages across quartiles (Q1–Q4) of the CHG index. Shown are (A) progression from CKM stage 0–1 to stage 2–3, and (B) progression from CKM stage 1–3 to stage 4. (C, D) Forest plots displaying Hazard Ratios (HRs) and 95% CIs for the association between CHG levels and CKM progression in different adjustment models. (C) Progression to stage 2–3; (D) Progression to stage 4. Analyses were performed comparing quartiles (Ref: Q1) and per 1 SD increase in CHG. (E, F) Fully adjusted restricted cubic spline (RCS) curves illustrating the dose-response relationship between the continuous CHG index and the risk of (E) progression to stage 2–3 and (F) progression to stage 4. Notes: Models were adjusted as follows: Model 1 adjusted for age, sex, and ethnicity. Model 2 adjusted for covariates in Model 1 plus socioeconomic status (Townsend Deprivation Index, household income, marital status, and employment status). Model 3 adjusted for covariates in Model 2 plus lifestyle factors (alcohol consumption, dietary habits, sleep pattern, physical activity, and smoking status). Model 4 adjusted for covariates in Model 3 plus metabolic and clinical profiles (body mass index, systolic blood pressure, HbA1c, LDL-cholesterol, and triglycerides)
Table 3.
Association Between CHG and Progression of Cardiovascular-Kidney-Metabolic Syndrome
| Model | Term | Progression from CKM stage 0-1 to 2-3 | Progression from CKM stage 1-3 to 4 | ||
|---|---|---|---|---|---|
| HR (95% CI) | p | HR (95% CI) | p | ||
| Model0 | Q1 | Ref | - | Ref | - |
| Q2 | 1.44 (1.32-1.57) | <0.001 | 1.05 (1.01-1.09) | 0.009 | |
| Q3 | 1.84 (1.68-2.02) | <0.001 | 1.20 (1.16-1.24) | <0.001 | |
| Q4 | 2.56 (2.26-2.91) | <0.001 | 1.58 (1.53-1.63) | <0.001 | |
| per 1 SD | 1.35 (1.31-1.40) | <0.001 | 1.22 (1.21-1.24) | <0.001 | |
| Model1 | Q1 | Ref | - | Ref | - |
| Q2 | 1.25 (1.15-1.36) | <0.001 | 0.97 (0.94-1.01) | 0.102 | |
| Q3 | 1.50 (1.37-1.65) | <0.001 | 1.07 (1.03-1.10) | <0.001 | |
| Q4 | 2.04 (1.79-2.32) | <0.001 | 1.38 (1.34-1.43) | <0.001 | |
| per 1 SD | 1.24 (1.19-1.29) | <0.001 | 1.18 (1.17-1.19) | <0.001 | |
| Model2 | Q1 | Ref | - | Ref | - |
| Q2 | 1.24 (1.14-1.35) | <0.001 | 0.97 (0.94-1.01) | 0.103 | |
| Q3 | 1.48 (1.34-1.62) | <0.001 | 1.06 (1.02-1.09) | <0.001 | |
| Q4 | 1.97 (1.73-2.24) | <0.001 | 1.35 (1.31-1.39) | <0.001 | |
| per 1 SD | 1.22 (1.18-1.27) | <0.001 | 1.16 (1.15-1.18) | <0.001 | |
| Model3 | Q1 | Ref | - | Ref | - |
| Q2 | 1.23 (1.13-1.34) | <0.001 | 0.97 (0.94-1.01) | 0.102 | |
| Q3 | 1.46 (1.32-1.60) | <0.001 | 1.05 (1.02-1.09) | 0.002 | |
| Q4 | 1.92 (1.69-2.19) | <0.001 | 1.33 (1.28-1.37) | <0.001 | |
| per 1 SD | 1.22 (1.17-1.26) | <0.001 | 1.15 (1.14-1.17) | <0.001 | |
| Model4 | Q1 | Ref | - | Ref | - |
| Q2 | 1.16 (1.06-1.27) | 0.001 | 0.94 (0.90-0.97) | <0.001 | |
| Q3 | 1.32 (1.18-1.49) | <0.001 | 0.98 (0.95-1.02) | 0.389 | |
| Q4 | 1.68 (1.43-1.97) | <0.001 | 1.13 (1.08-1.18) | <0.001 | |
| per 1 SD | 1.16 (1.11-1.23) | <0.001 | 1.06 (1.05-1.08) | <0.001 | |
Model 1: Adjusted for age, sex, and ethnicity. Model 2: Adjusted for covariates in Model 1 plus socioeconomic status (Townsend Deprivation Index, household income, marital status, and employment status). Model 3: Adjusted for covariates in Model 2 plus lifestyle factors (alcohol consumption, dietary habits, sleep pattern, physical activity, and smoking status). Model 4: Adjusted for covariates in Model 3 plus metabolic and clinical profiles (body mass index, systolic blood pressure, HbA1c, LDL-cholesterol, and triglycerides)
Prognostic significance of the CHG index for secondary prevention in CAD
In the cohort of 8,494 patients with established CAD, Kaplan-Meier analysis showed a significant increase in the cumulative incidence of MACCE across ascending CHG quartiles (Log-rank P < 0.001, Fig. 4 A).Consistently, the CHG index showed significant associations with a broad spectrum of other atherosclerosis-related events (Fig. 4 B–D). Furthermore, Supplementary Figure 3F confirms that when predicting the 2-year transition from Stage 4 to MACCE, the absolute risks calculated by the CHG index accurately reflect the real-world survival probabilities. In multivariable Cox proportional hazards models fully adjusted for potential confounders (Model 3), the CHG index remained significantly associated with recurrent adverse events (Fig. 4 E and Table 4). Participants in the highest quartile (Q4) had a higher risk of MACCE compared with those in the lowest quartile (Q1) (HR: 1.25; 95% CI: 1.03–1.50). Regarding specific endpoints, the association was most pronounced for the composite of MI and stroke, with an HR of 2.15 (95% CI: 1.12–4.13) for Q4 versus Q1. Restricted cubic spline analysis indicated a linear association between continuous CHG levels and the risk of MACCE (P for overall association = 0.002; P for non-linearity = 0.174), with the risk of events increasing monotonically with higher CHG levels (Fig. 4 F–I). As detailed in Supplementary Table 17, the CHG index demonstrated significantly superior performance in predicting the risk of MACCE at Stage 4 compared to TC, HDL-C, and GLU alone, which was consistently confirmed by all evaluated metrics, including AUC, NRI, and IDI.
Fig. 4.
Association of the CHG index with the risk of MACCE and secondary outcomes. (A–D) Kaplan-Meier cumulative incidence curves for (A) MACCE, (B) Death, MI, or Stroke, (C) MI or Stroke, and (D) Revascularization, stratified by quartiles (Q1–Q4) of the CHG index. (E) Forest plots showing Hazard Ratios (HRs) and 95% CIs for the associations between CHG levels and clinical outcomes across four adjustment models (Model 0–3). Analyses were performed comparing quartiles (Ref: Q1) and per 1 SD increase in CHG. (F–I) Fully adjusted restricted cubic spline (RCS) curves illustrating the dose-response relationship between the continuous CHG index and the risk of (F) MACCE, (G) Death, MI, or Stroke, (H) MI or Stroke, and (I) Revascularization. Blue lines indicate the unadjusted model; red lines indicate the fully adjusted model (Model 3). Shaded areas represent 95% CIs. Note: Model 1: Adjusted for age and sex. Model 2: Adjusted for covariates in Model 1 plus smoking status, comorbidities (diabetes mellitus, hypertension), and physical examination indicators (BMI, systolic blood pressure, heart rate). Model 3: Adjusted for covariates in Model 2 plus laboratory findings (ALT, creatinine, uric acid, hs-TnI, WBC, hemoglobin, total bilirubin, fasting blood glucose) and medication usage (aspirin, P2Y12 inhibitors, beta-blockers, ACEI, and statins)
Table 4.
Association Between CHG and Adverse Events among CAD patients
| Model | Term | MACCE | Death, MI, or Stroke | MI or Stroke | Revascularization | ||||
|---|---|---|---|---|---|---|---|---|---|
| HR (95% CI) | p | HR (95% CI) | p | HR (95% CI) | p | HR (95% CI) | p | ||
| Unadjusted | Q1 | Ref | - | Ref | - | Ref | - | Ref | - |
| Q2 | 1.05 (0.87-1.27) | 0.590 | 1.77 (1.14-2.75) | 0.011 | 2.00 (1.05-3.81) | 0.034 | 0.96 (0.79-1.18) | 0.725 | |
| Q3 | 1.45 (1.22-1.72) | <0.001 | 1.90 (1.23-2.92) | 0.004 | 2.49 (1.34-4.62) | 0.004 | 1.43 (1.19-1.72) | <0.001 | |
| Q4 | 1.65 (1.39-1.95) | <0.001 | 2.04 (1.33-3.12) | 0.001 | 2.98 (1.63-5.44) | <0.001 | 1.62 (1.35-1.94) | <0.001 | |
| per 1 SD | 1.23 (1.17-1.30) | <0.001 | 1.25 (1.11-1.42) | <0.001 | 1.40 (1.20-1.64) | <0.001 | 1.23 (1.16-1.31) | <0.001 | |
| Model 1 | Q1 | Ref | - | Ref | - | Ref | - | Ref | - |
| Q2 | 1.05 (0.87-1.26) | 0.628 | 1.75 (1.13-2.72) | 0.013 | 1.98 (1.04-3.76) | 0.037 | 0.96 (0.78-1.18) | 0.713 | |
| Q3 | 1.44 (1.21-1.71) | <0.001 | 1.88 (1.22-2.90) | 0.004 | 2.46 (1.33-4.57) | 0.004 | 1.42 (1.18-1.71) | <0.001 | |
| Q4 | 1.65 (1.39-1.95) | <0.001 | 2.14 (1.40-3.27) | <0.001 | 3.09 (1.69-5.63) | <0.001 | 1.61 (1.34-1.93) | <0.001 | |
| per 1 SD | 1.23 (1.17-1.30) | <0.001 | 1.29 (1.14-1.46) | <0.001 | 1.44 (1.22-1.69) | <0.001 | 1.23 (1.16-1.30) | <0.001 | |
| Model 2 | Q1 | Ref | - | Ref | - | Ref | - | Ref | - |
| Q2 | 1.01 (0.84-1.22) | 0.891 | 1.67 (1.07-2.60) | 0.023 | 1.94 (1.02-3.69) | 0.045 | 0.93 (0.76-1.14) | 0.498 | |
| Q3 | 1.34 (1.12-1.60) | 0.001 | 1.70 (1.09-2.64) | 0.018 | 2.26 (1.20-4.24) | 0.011 | 1.32 (1.10-1.60) | 0.004 | |
| Q4 | 1.43 (1.19-1.72) | <0.001 | 1.77 (1.13-2.79) | 0.013 | 2.52 (1.33-4.80) | 0.005 | 1.39 (1.14-1.69) | 0.001 | |
| per 1 SD | 1.17 (1.10-1.25) | <0.001 | 1.22 (1.06-1.40) | 0.006 | 1.34 (1.12-1.61) | 0.002 | 1.16 (1.09-1.24) | <0.001 | |
| Model 3 | Q1 | Ref | - | Ref | - | Ref | - | Ref | - |
| Q2 | 0.96 (0.80-1.16) | 0.702 | 1.57 (1.01-2.44) | 0.047 | 1.82 (0.96-3.48) | 0.069 | 0.89 (0.72-1.09) | 0.251 | |
| Q3 | 1.23 (1.03-1.47) | 0.024 | 1.54 (0.99-2.39) | 0.058 | 2.05 (1.08-3.87) | 0.027 | 1.23 (1.01-1.49) | 0.038 | |
| Q4 | 1.25 (1.03-1.50) | 0.021 | 1.45 (0.91-2.30) | 0.117 | 2.15 (1.12-4.13) | 0.021 | 1.23 (1.01-1.51) | 0.043 | |
| per 1 SD | 1.12 (1.05-1.19) | <0.001 | 1.14 (0.99-1.33) | 0.075 | 1.27 (1.05-1.53) | 0.012 | 1.11 (1.04-1.19) | 0.003 | |
Model 1: Adjusted for age and sex. Model 2: Adjusted for covariates in Model 1 plus smoking status, comorbidities (diabetes mellitus, hypertension), and physical examination indicators (BMI, systolic blood pressure, heart rate). Model 3: Adjusted for covariates in Model 2 plus laboratory findings (ALT, creatinine, uric acid, hs-TnI, WBC, hemoglobin, total bilirubin, fasting blood glucose) and medication usage (aspirin, P2Y12 inhibitors, beta-blockers, ACEI, and statins)
Effect modification by patient characteristics in the CHG–MACCE association
Participants were stratified into high- and low-CHG groups according to the median CHG value. Subgroup analyses were conducted to examine potential heterogeneity in the association between elevated CHG and MACCE risk across patient populations, with subgroups defined a priori based on clinically relevant characteristics (Fig. 5A). Statistically significant interactions were observed between CHG and hs-CRP (p for interaction = 0.047) and between CHG and HbA1c (p for interaction = 0.038), indicating that the association between elevated CHG and MACCE risk was stronger at lower hs-CRP levels and higher HbA1c levels. Elevated CHG was operationalized as a binary treatment, and a causal forest was fitted to estimate conditional (heterogeneous) treatment effects on MACCE risk. Variable importance—calculated as the normalized split frequency of each covariate across trees—identified hs-CRP and HbA1c as the most influential variables (Fig. 5B). Figure 5C displays a partial dependence plot (PDP) illustrating how the average predicted effect changes with a single feature while holding all others constant—an assumption that may not fully reflect physiological reality. The analysis suggested that the adverse effect of elevated CHG on MACCE risk increased with decreasing hs-CRP and increasing HbA1c levels. SHapley Additive exPlanations (SHAP) were further applied to interpret feature contributions to treatment-effect heterogeneity. Figure 5D presents a SHAP summary (beeswarm) plot showing the distribution of SHAP values for each feature, with point colors representing scaled feature values. Features are ordered by their mean absolute SHAP values, consistent with the SHAP importance ranking. Figure 5E illustrates the 2-year discriminative performance of CHG for MACCE across hs-CRP and HbA1c strata (time-dependent AUC). In descriptive comparisons, CHG achieved higher AUCs in the low hs-CRP and high HbA1c strata. Furthermore, the adverse effect of CHG was more pronounced in patients with hs-CRP < 2 mg/L and HbA1c > 6.5% (Fig. 5F and G).
Fig. 5.
Heterogeneity of the association between CHG and MACCE. (A) Subgroup Analyses for the Association Between CHG Index and Risk of MACCE. (B) Variable importance from causal forest analysis. (C) Partial dependence plots of key covariates. (D) SHAP summary plot for feature contributions. (E) ROC curves showing discriminatory performance stratified by hs-CRP and HbA1c. (F) Joint stratification by hs-CRP and HbA1c with corresponding hazard ratios. (G) ROC curves showing the discriminatory performance of CHG, based on joint stratification by hs-CRP and HbA1c
Discussion
In this study, using data from two large-scale prospective cohorts, the UK Biobank and a clinical cohort from Beijing Anzhen Hospital, we investigated the association of the CHG index with the incidence, progression, and prognosis of CKM syndrome. Our principal findings are fourfold. First, in the general population, elevated baseline CHG levels are associated with a significant, graded increase in the risk of incident T2DM, CKD, and CVD; Second, the CHG index robustly predicted both the onset of CKM syndrome and its subsequent progression across stages; Third, even among patients with established coronary artery disease (CKM stage 4), an elevated CHG index remained significantly associated with a higher risk of MACCE; Finally, subgroup analyses and causal forest modeling indicated that while the adverse impact of CHG on MACCE was consistent across most clinical profiles, the magnitude of this effect was significantly modified by baseline hs-CRP and HbA1c levels.
Consistency with existing evidence and advancement toward a life-course framework
Our findings confirm the CHG index as a reliable metabolic metric, aligning with earlier evidence in primary prevention. While the index was initially proposed for T2DM diagnosis (AUC 0.864) [17], our analysis of 370,916 UK Biobank participants provides robust validation, showing an elevated risk of incident T2DM in the highest quartile (HR 2.49; 95% CI 2.31–2.70). These results also support observations by Mo et al. and Xiao et al., who linked elevated CHG to incident CVD and stroke in adults with normal glucose regulation [18, 25]. By documenting 39,893 incident CVD events, our study reinforces the index’s sensitivity to early metabolic dysregulation across a much larger, multi-ethnic population. Beyond early-stage onset, our study explores the prognostic value of CHG in advanced clinical settings—a domain previously constrained by retrospective or small-scale designs [26, 27]. Although prior work identified CHG as a predictor of kidney decline and post-stroke recovery, data on established coronary artery disease (CKM Stage 4) remained sparse [28–31]. Our prospective follow-up of 8,494 CAD patients addresses this gap. We observed that elevated CHG independently predicts MACCE (HR 1.25; 95% CI 1.03–1.50), with a particularly strong association with the composite of MI and stroke (HR 2.15; 95% CI 1.12–4.13). This indicates that the metabolic risk captured by CHG persists as a potent driver of adverse outcomes even in patients already receiving guideline-directed medical therapy. To avoid multicollinearity among the adjusted metabolic indicators, the Variance Inflation Factor (VIF) values are presented in Supplementary Table 15, demonstrating that the results of the multivariable adjustment are robust.
The primary advancement of this study lies in its integrated, life-course perspective on CKM syndrome. Most existing research has treated CKM components as isolated disease events [17, 25]. By contrast, we tracked the transition across the entire disease spectrum, demonstrating that high CHG drives the progression from early CKM (Stages 0–1) to intermediate (Stages 2–3) and finally to clinical CVD (Stage 4). Furthermore, to move beyond the limitations of traditional linear modeling, we employed a causal forest algorithm to resolve the high-dimensional interactions between baseline clinical profiles and metabolic risk [21]. This approach allowed us to identify specific effect modifiers, such as hs-CRP and HbA1c, offering a precision-medicine framework for metabolic risk stratification that was previously uncaptured.
Furthermore, the clinical relevance of our findings is particularly underscored by the recent COVID-19 pandemic. Emerging evidence suggests an increased prevalence and accelerated progression of CKM syndrome during and after the pandemic, largely driven by detrimental lifestyle shifts, delayed routine preventive care, and the direct cardiometabolic impact of SARS-CoV-2 infection [32–34]. Consequently, identifying high-risk individuals using accessible and robust tools like the CHG index has become even more imperative in the post-pandemic era.
Mechanistic basis for the CHG index
The robust performance of the CHG index in predicting CKM progression and prognosis stems from its ability to integrate multiple interconnected pathological pathways. Rather than simply aggregating individual risks, the CHG index captures the biological synergy between TC and FBG [15, 16]. The co-elevation of these parameters exacerbates the production of reactive oxygen species (ROS) and triggers the accumulation of advanced glycation end products (AGEs) [8, 35–38]. This synergistic effect accelerates vascular endothelial injury and target organ damage, serving as a fundamental driver for both microvascular (CKD) and macrovascular (CVD) complications [39–41]. Furthermore, the inclusion of HDL-C as the denominator makes the index highly sensitive to the deficiency of vascular protection. Beyond its classical role in reverse cholesterol transport, HDL-C exerts potent anti-inflammatory and antioxidant effects [42, 43]. Specifically, it inhibits LDL oxidation via paraoxonase 1 (PON1) activity and downregulates the expression of key adhesion molecules, such as VCAM-1 and ICAM-1 [44–47]. As a validated surrogate for insulin resistance (IR), the CHG index reflects a foundational pathology common across the CKM spectrum. IR facilitates a state of chronic inflammation and endothelial dysfunction, which together impair the cardio-renal axis and drive the transition from subclinical metabolic derangement to overt clinical events [48–50]. By integrating these diverse yet converging mechanisms, the CHG index provides a more comprehensive reflection of the underlying metabolic burden [51].
Non-linear Association and Metabolic Thresholds
The RCS analysis revealed significant non-linear associations between the CHG index and incident T2DM, CVD, as well as stage-to-stage progression within the general population (all Nonlinear P < 0.001). Similar findings were reported by Zhang et al., who identified a CHG threshold of 5.24 as a critical point for sharply elevated T2DM risk [52]. Such trajectories highlight a metabolic threshold in the natural history of CKM syndrome. At lower CHG levels, the relatively flat risk profile suggests a period of metabolic compensation, where adipose tissue effectively buffers the lipid load. However, once these biological thresholds are breached, the steep escalation in risk likely reflects the onset of glucolipotoxicity and ectopic lipid deposition [53].
However, the sharp escalation in risk at higher levels suggests a transition into pathophysiological decompensation. Once biological thresholds are breached, the combined burden of excess lipid and glucose likely trigger target organ damage and accelerate CKM stage deterioration [54]. Notably, this non-linear pattern was absent in the Anzhen Hospital cohort, where the risk of MACCE in patients with established CAD increased linearly and monotonically with higher CHG levels (Non-linear P = 0.174). This contrast suggests that in advanced CKM Stage 4, the protective metabolic buffering capacity is largely exhausted, making the patient significantly more vulnerable to even small increases in metabolic burden [51]. Clinically, these findings underscore the necessity of identifying and treating individuals before their metabolic risk profile reaches the steep portion of the risk curve.
Heterogeneity of the CHG-MACCE association
In our high-risk cohort of 8,494 patients with established CAD, the CHG index remained a significant predictor of MACCE (Q4 vs. Q1 HR 1.25; 95% CI 1.03–1.50). To move beyond traditional categorical subgroups, we employed a causal forest model to resolve the multidimensional interactions between a patient’s clinical profile and their metabolic risk. This machine learning approach identified hs-CRP and HbA1c as the primary modifiers of CHG’s prognostic value.
Causal forest analysis revealed that the prognostic impact of CHG was more pronounced in patients with low hs-CRP or high HbA1c. In low-inflammatory states, CHG primarily reflects underlying metabolic disturbances and vascular dysfunction, enhancing its predictive utility [55, 56]. Conversely, at higher hs-CRP levels, inflammation functions as a dominant risk driver, limiting the incremental contribution of CHG [57]. This attenuation may result from inflammatory remodeling that converts HDL into pro-inflammatory particles, alongside a negative additive interaction (RERI < 0) where high inflammation reduces the relative risk contribution of dyslipidemia [58]. We also found that the association between CHG and MACCE was more pronounced in patients with higher HbA1c. In this subset, fasting glucose levels more accurately represent chronic glucotoxicity rather than transient fluctuations [59]. Furthermore, patients with elevated HbA1c frequently exhibit a more severe atherogenic profile, often characterized by higher total cholesterol and dysfunctional HDL [42, 43, 60]. These pathological features align closely with the biochemical components of the CHG index, thereby reinforcing its predictive performance in cardiovascular risk assessment.
Limitations
Several limitations of our study should be acknowledged. First, despite the prospective design and the application of advanced causal machine learning techniques to mitigate bias, the observational nature of the study precludes definitive causal inference. Although we adjusted for a comprehensive range of potential confounders, the possibility of residual confounding from unmeasured variables (e.g., detailed dietary patterns or distinct socioeconomic shifts during follow-up) remains. Second, the calculation of the CHG index was based on a single baseline measurement. Consequently, we could not account for longitudinal fluctuations in lipid or glucose levels, nor could we evaluate the impact of trajectory changes in the CHG index following the initiation of secondary prevention therapies. This reliance on a single time point may introduce regression dilution bias, potentially underestimating the true strength of the associations. Notably, our analysis lacked specific data regarding the use of sodium-glucose cotransporter-2 inhibitors (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP1 RA). Given their potent metabolic effects and direct cardiovascular and renal protection, the inability to adjust for these therapies represents a potential source of treatment-related confounding in the context of modern CKM management. Then, the definition of CKM stages and the identification of clinical events (e.g., CVD or CKD) relied on ICD codes and electronic health records. While these methods are standardized, a degree of misclassification or underreporting of subclinical cases is inherent in large-scale database research. The approximately 10% loss to follow-up in our external validation cohort (Beijing Anzhen Hospital) may introduce selection bias, as these patients might have differed in their baseline risk profiles or undocumented MACCE events, potentially influencing the validation performance. Finally, the CHG index is a simplified metric that cannot offer incremental predictive value over established multivariable algorithms like the PREVENT or Framingham risk scores. However, its reliance on standard, universally available laboratory panels makes it a valuable, cost-effective tool for rapid preliminary screening, especially in primary care or resource-constrained clinical environments.
Conclusion
Our study establishes the CHG index as a versatile predictor that spans the entire clinical spectrum of CKM syndrome. In the general population, the index not only identifies the initial onset of T2DM, CKD, and CVD but also effectively tracks the progression between subclinical and overt disease stages. This predictive utility extends into advanced clinical settings, where the CHG index remains a potent prognostic marker for MACCE and recurrent events in patients with established coronary artery disease. The study further highlights that the index’s performance is notably modified by baseline inflammatory and glycemic status, specifically hs-CRP and HbA1c.
Supplementary Information
Acknowledgements
We are grateful to the participants and investigators of the UK Biobank for making this research possible. This research has been conducted using the UK Biobank Resource under Application Number 103736. We also wish to extend our special thanks to our colleagues at the Department of Cardiology, Beijing Anzhen Hospital, for their invaluable efforts and assistance during the data application process.
Abbreviations
- AGEs
Advanced glycation end-products
- AHA
American heart association
- ANOVA
Analysis of variance
- AUC
Area under the curve
- CAD
Coronary artery disease
- CATEs
Conditional average treatment effects
- CHG
Cholesterol, High-density lipoprotein, and Glucose
- CI
Confidence interval
- CIF
Cumulative incidence function
- CKD
Chronic kidney disease
- CKM
Cardiovascular-kidney-metabolic
- CVD
Cardiovascular disease
- eGFR
Estimated glomerular filtration rate
- FBG
Fasting blood glucose
- GDMT
Guideline-directed medical therapy
- Glu
Glucose
- HbA1c
Glycated hemoglobin
- HDL-C
High-density lipoprotein cholesterol
- HF
Heart failure
- HR
Hazard ratio
- hs-CRP
High-sensitivity C-reactive protein
- ICD
International classification of diseases
- IQR
Interquartile range
- IR
Insulin resistance
- LDL-C
Low-density lipoprotein cholesterol
- MACCE
Major adverse cardiovascular and cerebrovascular events
- MI
Myocardial infarction
- PCI
Percutaneous coronary intervention
- PDP
Partial dependence plot
- RCS
Restricted cubic spline
- ROS
Reactive oxygen species
- SD
Standard deviation
- SHAP
SHapley additive exPlanations
- SMD
Standardized mean difference
- T2DM
Type 2 diabetes mellitus
- TC
Total cholesterol
- TG
Triglycerides
Author contributions
Zehao Zhao and Yingxun Bai: Conceptualization, Formal analysis, Writing – original draft. Changsheng Ma: Data curation. Tianyi Wang: Formal analysis. Libo Zhao: Formal analysis. Xiliang Zhao: Formal analysis. Quan Li: Formal analysis, Visualization. Yaodong Ding: Formal analysis. Yang Zhang: Formal analysis. Li Lin: Writing – review & editing. Wenjie Wang: Writing – review & editing. Jiayi Han: Writing – review & editing. Yong Zeng: Conceptualization, Formal analysis, Writing – review & editing.
Funding
This work was supported by the National Natural Science Foundation of China (General Program) [Grant No. 82270345], the Capital Health Research and Development Special Project of Beijing Municipal Health Commission (Key Research Project) [Grant No. Shoufa 2024-1-2061], and the High-level Research Special Fund (RCT Project) of Beijing Anzhen Hospital [Grant No. 2024AZB1011].
Data availability
Data supporting the findings of this study from the UK Biobank team (http://www.ukbiobank.ac.uk/) and the Beijing Anzhen Hospital. The data and methods that support the findings of this study are available from the corresponding author on reasonable request and with approval from the Information Center and Ethics Committee of Beijing Anzhen Hospital.
Declarations
Ethics approval and consent to participate
The UK Biobank study received ethical approval from the North West Multi-centre Research Ethics Committee (MREC). This specific research was conducted using the UK Biobank Resource under Application Number 103736. The Anzhen cohort study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Beijing Anzhen Hospital, Capital Medical University (approval No. KS2023081). Written informed consent was obtained from all participants prior to participation.
Consent for publication
All participants provided informed electronic consent at recruitment.
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.
Zehao Zhao and Yingxun Bai contributed equally to this work.
References
- 1.Ndumele CE, Neeland IJ, Tuttle KR, Chow SL, Mathew RO, Khan SS, et al. A synopsis of the evidence for the science and clinical management of cardiovascular-kidney-metabolic (CKM) syndrome: a scientific statement from the American Heart Association. Circulation. 2023;148(20):1636–64. [DOI] [PubMed] [Google Scholar]
- 2.Khan SS, Coresh J, Pencina MJ, Ndumele CE, Rangaswami J, Chow SL, et al. Novel prediction equations for absolute risk assessment of total cardiovascular disease incorporating cardiovascular-kidney-metabolic health: a scientific statement from the American Heart Association. Circulation. 2023;148(24):1982–2004. [DOI] [PubMed] [Google Scholar]
- 3.Claudel SE, Verma A. Cardiovascular-kidney-metabolic syndrome: a step toward multidisciplinary and inclusive care. Cell Metab. 2023;35(12):2104–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Zhang C, Chen Y, Wang T, Peng W, Ning M, Liu Y. Inflammation and abnormal glucose-lipid metabolism are critical for population stratification toward progressive frailty in cardiovascular-kidney-metabolic syndrome. Diabetes Res Clin Pract. 2025;230:112989. [DOI] [PubMed] [Google Scholar]
- 5.Larkin H. Here’s what to know about cardiovascular-kidney-metabolic syndrome, newly defined by the AHA. JAMA. 2023;330(21):2042–3. [DOI] [PubMed] [Google Scholar]
- 6.Li W, Shen C, Kong W, Zhou X, Fan H, Zhang Y, et al. Association between the triglyceride glucose-body mass index and future cardiovascular disease risk in a population with cardiovascular-kidney-metabolic syndrome stage 0-3: a nationwide prospective cohort study. Cardiovasc Diabetol. 2024;23(1):292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Sebastian SA, Padda I, Johal G. Cardiovascular-kidney-metabolic (CKM) syndrome: a state-of-the-art review. Curr Probl Cardiol. 2024;49(2):102344. [DOI] [PubMed] [Google Scholar]
- 8.Grumbach IM, Nguyen EK. Metabolic stress. Arterioscler Thromb Vasc Biol. 2019;39(6):991–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Global burden of 292 causes of death in 204 countries and territories and 660 subnational locations, 1990-2023: a systematic analysis for the Global Burden of Disease Study 2023. Lancet. 2025;406(10513):1811-72. [DOI] [PMC free article] [PubMed]
- 10.Claudel SE, Schmidt IM, Waikar SS, Verma A. Cumulative incidence of mortality associated with cardiovascular-kidney-metabolic (CKM) syndrome. J Am Soc Nephrol. 2025;36(7):1343–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Ji H, Sabanayagam C, Matsushita K, Cheng CY, Rim TH, Sheng B, et al. Sex Differences in Cardiovascular-Kidney-Metabolic Syndrome: 30-Year US Trends and Mortality Risks-Brief Report. Arterioscler Thromb Vasc Biol. 2025;45(1):157–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ostrominski JW, Arnold SV, Butler J, Fonarow GC, Hirsch JS, Palli SR, et al. Prevalence and Overlap of Cardiac, Renal, and Metabolic Conditions in US Adults, 1999–2020. JAMA Cardiol. 2023;8(11):1050–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Tsai MK, Kao JT, Wong CS, Liao CT, Lo WC, Chien KL, et al. Cardiovascular-kidney-metabolic syndrome and all-cause and cardiovascular mortality: A retrospective cohort study. PLoS Med. 2025;22(6):e1004629. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Li N, Li Y, Cui L, Shu R, Song H, Wang J, et al. Association between different stages of cardiovascular-kidney-metabolic syndrome and the risk of all-cause mortality. Atherosclerosis. 2024;397:118585. [DOI] [PubMed] [Google Scholar]
- 15.Zhu H, Liu L, Yang S, Fu Y, Pan Y, Fu Q, et al. Association of cholesterol, high-density lipoprotein and glucose (CHG) index with mortality risk in metabolic dysfunction-associated steatotic liver disease (MASLD) adults: results from two prospective cohorts. Cardiovasc Diabetol. 2026;25(1):23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Wei S, Yu S, Qian C, Xu Z, Jia Y, Chen B. Exploring the U-shaped nonlinear relationship of CHG index with metabolic syndrome and mortality risks in metabolic syndrome patients. Lipids Health Dis. 2025;24(1):294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Mansoori A, Nosrati M, Dorchin M, Mohammadyari F, Derakhshan-Nezhad E, Ferns G, et al. A novel index for diagnosis of type 2 diabetes mellitus: Cholesterol, High density lipoprotein, and Glucose (CHG) index. J Diabetes Investig. 2025;16(2):309–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Mo D, Zhang P, Zhang M, Dai H, Guan J. Cholesterol, high-density lipoprotein, and glucose index versus triglyceride-glucose index in predicting cardiovascular disease risk: a cohort study. Cardiovasc Diabetol. 2025;24(1):116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Çatak M, Konuk ŞG, Hepsen S. The cholesterol-HDL-glucose (CHG) index and traditional adiposity markers in predicting diabetic retinopathy and nephropathy. J Diabetes Investig. 2025;16(8):1487–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Zhang Y, Zhu Y, He Y, Yan Y, Zhao Q, Du Y. Association of cumulative cholesterol-HDL-glucose index with blood pressure changes and risk of new-onset hypertension in middle-aged and older adults: a cohort study. Cardiovasc Diabetol. 2026. 10.1186/s12933-026-03081-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Athey S, Imbens G. Recursive partitioning for heterogeneous causal effects. Proc Natl Acad Sci U S A. 2016;113(27):7353–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Chinese Society of Cardiology CMAEBoCJoC. Guidelines for percutaneous coronary intervention (2025). Zhonghua Xin Xue Guan Bing Za Zhi. 2025;53(7):717–45. [DOI] [PubMed] [Google Scholar]
- 23.Aggarwal R, Ostrominski JW, Vaduganathan M. Prevalence of Cardiovascular-Kidney-Metabolic Syndrome stages in US adults, 2011-2020. JAMA. 2024;331(21):1858–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Khan SS, Matsushita K, Sang Y, Ballew SH, Grams ME, Surapaneni A, et al. Development and validation of the American Heart Association’s PREVENT equations. Circulation. 2024;149(6):430–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Xiao Q, Huang W, Zhuang Y, Chen Y, Peng G, Li Y. Evaluating the role of CHG index in predicting stroke risk among adults with varying glucose regulation. Aging Clin Exp Res. 2025;37(1):334. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Tian R, Tian X, Chen S, Li X, Zheng M, Wu S, et al. Association of cumulative exposure to cholesterol, high-density lipoprotein, and glucose index with the risk of cardiovascular disease and all-cause mortality: A longitudinal cohort study. Diabetes Obes Metab. 2026;28(2):1453–62. [DOI] [PubMed] [Google Scholar]
- 27.Ke L, Li Y, Jiang S, Xing W, Zhao L. Associations between the CHG index, its modified versions and incident stroke in patients with early-stage CKM syndrome: a nationwide cohort study. Eur J Med Res. 2026. [DOI] [PMC free article] [PubMed]
- 28.Yang C, Ma M, Jiang X, Zhang L, Zhang Y, Zhou M, et al. Association of the CHG index with 90-day functional outcomes and mortality in acute ischemic stroke after endovascular therapy: a retrospective study. J Stroke Cerebrovasc Dis. 2025;34(12):108502. [DOI] [PubMed] [Google Scholar]
- 29.Luwen H, Linlin L, Ming Y, Lei X. A comparative analysis of the cholesterol-high-density lipoprotein-glucose index and the triglyceride-glucose index in predicting in-hospital mortality in critically ill ischemic stroke patients. Front Neurol. 2025;16:1664891. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Liu H, Tang G, Yuan Y, Gu P, Wang A, Zhao S, et al. Cholesterol, high-density lipoprotein, and glucose index is associated with renal survival in patients with IgA nephropathy. Front Endocrinol (Lausanne). 2025;16:1643360. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Dong M, Chen X, Liu J, Li C. Non-conventional lipid parameters predict the risk of rapid kidney function decline in patients with cardiovascular-kidney-metabolic syndrome: the first cohort study evidence from CHARLS. Exp Gerontol. 2025;210:112892. [DOI] [PubMed] [Google Scholar]
- 32.Trimarco V, Izzo R, Pacella D, Manzi VirginiaM., Trama U, Lembo M, et al. Increased prevalence of cardiovascular-kidney-metabolic syndrome during COVID-19: a propensity score-matched study. Diabetes Res Clin Pract. 2024;218:111926. [DOI] [PubMed] [Google Scholar]
- 33.Zuin M, Rigatelli G, Battisti V, Costola G, Roncon L, Bilato C. Increased risk of acute myocardial infarction after COVID-19 recovery: a systematic review and meta-analysis. Int J Cardiol. 2023;372:138–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Xie Y, Xu E, Bowe B, Al-Aly Z. Long-term cardiovascular outcomes of COVID-19. Nat Med. 2022;28(3):583–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Zhang B, Pan C, Feng C, Yan C, Yu Y, Chen Z, et al. Role of mitochondrial reactive oxygen species in homeostasis regulation. Redox Rep. 2022;27(1):45–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Habtemariam S. Modulation of reactive oxygen species in health and disease. Antioxidants. 2019. 10.3390/antiox8110513. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Ayer A, Fazakerley DJ, James DE, Stocker R. The role of mitochondrial reactive oxygen species in insulin resistance. Free Radic Biol Med. 2022;179:339–62. [DOI] [PubMed] [Google Scholar]
- 38.Zawada A, Machowiak A, Rychter AM, Ratajczak AE, Szymczak-Tomczak A, Dobrowolska A, et al. Accumulation of advanced glycation end-products in the body and dietary habits. Nutrients. 2022. 10.3390/nu14193982. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Nowak WN, Deng J, Ruan XZ, Xu Q. Reactive Oxygen Species generation and atherosclerosis. Arterioscler Thromb Vasc Biol. 2017;37(5):e41–52. [DOI] [PubMed] [Google Scholar]
- 40.Martens MD, Holody CD, Wells L, Silver HL, Morales-Llamas DY, Du WW, et al. Reactive Oxygen Species modulator 1 plays an obligate role in cardiomyocyte hypertrophy. Circ Res. 2024;134(1):114–6. [DOI] [PubMed] [Google Scholar]
- 41.Wells AT, Bossardi Ramos R, Shen MM, Binrouf RH, Swinegar AE, Lennartz MR. Identification of myeloid Protein Kinase C Epsilon as a novel atheroprotective gene. Arterioscler Thromb Vasc Biol. 2025;45(9):e392–411. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Rohatgi A, Westerterp M, von Eckardstein A, Remaley A, Rye KA. HDL in the 21st century: a multifunctional roadmap for future HDL research. Circulation. 2021;143(23):2293–309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.von Eckardstein A, Nordestgaard BG, Remaley AT, Catapano AL. High-density lipoprotein revisited: biological functions and clinical relevance. Eur Heart J. 2023;44(16):1394–407. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Kotur-Stevuljević J, Vekić J, Stefanović A, Zeljković A, Ninić A, Ivanišević J, et al. Paraoxonase 1 and atherosclerosis-related diseases. Biofactors. 2020;46(2):193–205. [DOI] [PubMed] [Google Scholar]
- 45.Durrington PN, Mackness B, Mackness MI. Paraoxonase and atherosclerosis. Arterioscler Thromb Vasc Biol. 2001;21(4):473–80. [DOI] [PubMed] [Google Scholar]
- 46.Damiza-Detmer A, Pawełczyk M, Głąbiński A. Protective role of high-density lipoprotein in Multiple Sclerosis. Antioxidants. 2024. 10.3390/antiox13111276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Yang L, Wang Y, Xu Y, Li K, Yin R, Zhang L, et al. ANGPTL3 is a novel HDL component that regulates HDL function. J Transl Med. 2024;22(1):263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Di Pino A, DeFronzo RA. Insulin resistance and atherosclerosis: implications for insulin-sensitizing agents. Endocr Rev. 2019;40(6):1447–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Hill MA, Yang Y, Zhang L, Sun Z, Jia G, Parrish AR, et al. Insulin resistance, cardiovascular stiffening and cardiovascular disease. Metabolism. 2021;119:154766. [DOI] [PubMed] [Google Scholar]
- 50.Matsuda M. Measuring and estimating insulin resistance in clinical and research settings. Nutr Metab Cardiovasc Dis. 2010;20(2):79–86. [DOI] [PubMed] [Google Scholar]
- 51.Samuel VT, Shulman GI. Mechanisms for insulin resistance: common threads and missing links. Cell. 2012;148(5):852–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Zhang Z, Weng S, Cai D, Zhao L, Chen M. The relationship between cholesterol, high-density lipoprotein, and glucose index and hypertension: a study based on two national cohorts. Sci Prog. 2025;108(4):368504251396781. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Borén J, Taskinen MR, Olofsson SO, Levin M. Ectopic lipid storage and insulin resistance: a harmful relationship. J Intern Med. 2013;274(1):25–40. [DOI] [PubMed] [Google Scholar]
- 54.Han E, Han KD, Lee BW, Kang ES, Cha BS, Ko SH, et al. Severe Hypoglycemia Increases Dementia Risk and Related Mortality: A Nationwide, Population-based Cohort Study. J Clin Endocrinol Metab. 2022;107(5):e1976–86. [DOI] [PubMed] [Google Scholar]
- 55.Ridker PM, Rifai N, Rose L, Buring JE, Cook NR. Comparison of C-reactive protein and low-density lipoprotein cholesterol levels in the prediction of first cardiovascular events. N Engl J Med. 2002;347(20):1557–65. [DOI] [PubMed] [Google Scholar]
- 56.Kaptoge S, Di Angelantonio E, Pennells L, Wood AM, White IR, Gao P, et al. C-reactive protein, fibrinogen, and cardiovascular disease prediction. N Engl J Med. 2012;367(14):1310–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57.Bafei SEC, Zhao X, Chen C, Sun J, Zhuang Q, Lu X, et al. Interactive effect of increased high sensitive C-reactive protein and dyslipidemia on cardiovascular diseases: a 12-year prospective cohort study. Lipids Health Dis. 2023;22(1):95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58.Vaisar T, Tang C, Babenko I, Hutchins P, Wimberger J, Suffredini AF, et al. Inflammatory remodeling of the HDL proteome impairs cholesterol efflux capacity. J Lipid Res. 2015;56(8):1519–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Campos C. Chronic hyperglycemia and glucose toxicity: pathology and clinical sequelae. Postgrad Med. 2012;124(6):90–7. [DOI] [PubMed] [Google Scholar]
- 60.Borén J, Chapman MJ, Krauss RM, Packard CJ, Bentzon JF, Binder CJ, et al. Low-density lipoproteins cause atherosclerotic cardiovascular disease: pathophysiological, genetic, and therapeutic insights: a consensus statement from the European Atherosclerosis Society Consensus Panel. Eur Heart J. 2020;41(24):2313–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data supporting the findings of this study from the UK Biobank team (http://www.ukbiobank.ac.uk/) and the Beijing Anzhen Hospital. The data and methods that support the findings of this study are available from the corresponding author on reasonable request and with approval from the Information Center and Ethics Committee of Beijing Anzhen Hospital.