The cholesterol, high-density lipoprotein, and glucose index and its derivatives related to obesity and inflammation as novel metabolic markers for predicting new-onset cardiovascular-liver-metabolic multimorbidity: A prospective cohort study

Zhonghua Sun; Aikeliyaer Ainiwaer; Yameng Xu; Ge Zhang; Xiangyu Sun; Xinliang Peng; Yuanhui Dai; Kaisaierjiang Kadier

doi:10.1111/jdi.70462

Article · 2026 Paper GEO V2

The cholesterol, high-density lipoprotein, and glucose index and its derivatives related to obesity and inflammation as novel metabolic markers for predicting new-onset cardiovascular-liver-metabolic multimorbidity: A prospective cohort study

Journal of Diabetes Investigation

Full Authors

Zhonghua Sun
Aikeliyaer Ainiwaer
Yameng Xu
Ge Zhang
Xiangyu Sun
Xinliang Peng
Yuanhui Dai
Kaisaierjiang Kadier

DOI / publisher Google Scholar search Cite

Official Abstract

Background

Cardiovascular‐liver‐metabolic multimorbidity (CLMM), defined as the co‐occurrence of at least two of cardiovascular disease, type 2 diabetes, and metabolic dysfunction‐associated steatotic liver disease, is an increasing health burden. We evaluated the cholesterol, high‐density lipoprotein, and glucose (CHG) index and its obesity‐ and inflammation‐derived variants for predicting incident CLMM.

Methods

This prospective cohort included 6,534 adults aged ≥45 years from the China Health and Retirement Longitudinal Study. CHG, eight obesity‐derived indices, and CHG‐CRP were assessed, with the triglyceride‐glucose index and TG/HDL‐C ratio as comparators. Associations with incident CLMM were evaluated using logistic regression, restricted cubic splines, and threshold analysis. Predictive performance was compared using receiver operating characteristic analysis, the integrated discrimination improvement (IDI), and the net reclassification improvement (NRI), with additional joint‐effect, interaction, subgroup, and sensitivity analyses.

Results

At the 2015 follow‐up, 406 participants (6.2%) met the criteria for CLMM. Higher CHG was linearly associated with incident CLMM (highest vs lowest quartile: OR 1.64, 95% confidence interval [CI] 1.22–2.22). Obesity‐derived variants showed stronger associations, particularly CHG‐BMI (OR 3.13, 95% CI 2.21–4.43), which also had the highest discrimination (AUC 0.643, 95% CI 0.616–0.669) and continuous NRI (34.08%). Participants with both high CHG and high BMI had the greatest CLMM risk (OR 2.59, 95% CI 1.94–3.51). Results were robust in sensitivity analyses.

Conclusions

The CHG index and its obesity‐derived variants, particularly CHG‐BMI, are accessible and inexpensive candidate metabolic markers for predicting new‐onset CLMM. CHG‐BMI showed the highest apparent discrimination among the evaluated indices, although its AUC was modest. External population validation is needed.

Text reproduced verbatim from PMC (source); DOI; CC-BY-NC-4.0.

Research Question

Are baseline CHG and its obesity- or inflammation-derived indices associated with new-onset cardiovascular–liver–metabolic multimorbidity, and how well do they discriminate follow-up cases within CHARLS?

Author Evidence Summary

This observational study related baseline metabolic indices to a later binary multimorbidity outcome in Chinese middle-aged and older adults. Multimorbidity required at least two component conditions and included an LAP-based suspected MASLD phenotype rather than imaging-confirmed liver disease. CHG-BMI had a stronger quartile association and a higher apparent AUC than the other evaluated indices. The findings support candidate-marker research; they do not establish causation, independent validation, an actionable clinical threshold, or benefit from modifying CHG-BMI.

Cite this paper

Plain citation

Zhonghua Sun, Aikeliyaer Ainiwaer, Yameng Xu, Ge Zhang, Xiangyu Sun, Xinliang Peng, Yuanhui Dai, Kaisaierjiang Kadier. The cholesterol, high-density lipoprotein, and glucose index and its derivatives related to obesity and inflammation as novel metabolic markers for predicting new-onset cardiovascular-liver-metabolic multimorbidity: A prospective cohort study. Journal of Diabetes Investigation. 2026. https://doi.org/10.1111/jdi.70462

Author-controlled citation; verify against the publisher version of record.

BibTeX RIS / EndNote CSL JSON DOI

Research & review context

Research fields: Cardiometabolic epidemiology
Metabolic biomarker research
Review topics: Multimorbidity risk-marker evaluation
Obesity-derived biochemical indices
Scientific role: Prospective observational association study with apparent discrimination analysis, requiring independent validation.
Evidence position: Population-level candidate-marker evidence with a surrogate liver-disease component; causal and clinical-utility claims remain unestablished.
Related concepts: CHG-BMI
CHARLS
LAP-based suspected MASLD
Possible citation contexts: Reviews comparing metabolic indices for incident, surrogate-defined cardiometabolic multimorbidity.
Methodological discussions separating observational odds ratios and apparent AUCs from validated clinical prediction.

Possible literature-review contexts, not evidence of existing citations. Read the findings and Evidence Scope before citing the original article.

Questions & evidence · Evidence limits · Original article & citation downloads

Evidence Snapshot

CHG-derived metabolic indices and incident cardiovascular–liver–metabolic multimorbidity in CHARLS

In a prospective CHARLS analysis, baseline CHG and obesity-derived variants were associated with subsequently meeting a surrogate-defined multimorbidity endpoint. CHG-BMI showed the largest observed AUC among the evaluated indices, but discrimination was modest and assessed in the analysis sample without validation.

Unique total: n=6,534
CHARLS analytic cohort: n=6,534

Key Findings

KF1 A prospective CHARLS analysis identified incident, surrogate-defined cardiovascular–liver–metabolic multimorbidity at follow-up.

Chinese community-dwelling adults without baseline CLMM; component conditions were evaluated at baseline and the follow-up wave, rather than using exact onset dates.

Evidence for KF1
Measure	Value
Analytic participants	6,534 adults aged 45 years or older
Exposure and outcome waves	2011 baseline and 2015 follow-up
Incident CLMM	406 participants; 6.2%
Outcome definition	At least two of CVD, type 2 diabetes, and LAP-based suspected MASLD

Source locator: Methods: Study population and Outcome assessment (jdi70462-sec-0007; jdi70462-sec-0009); Results 3.1 (jdi70462-sec-0013); Figure 1; Table 1

Permalink to KF1

KF2 Higher baseline CHG was associated with higher odds of meeting the CLMM criteria at follow-up.

Highest versus lowest CHG quartile in the fully adjusted logistic model; these are odds ratios for a binary follow-up outcome, not hazard ratios or treatment effects.

Evidence for KF2
Measure	Value
CHG Q4 versus Q1 adjusted OR	1.64; 95% CI 1.22–2.22
Adjusted covariates	Age, sex, education, residence, marital status, smoking, drinking, hypertension, and CKD

Source locator: Results 3.2 (jdi70462-sec-0014); Table 2 (jdi70462-tbl-0002), CHG Q4, Model 2

Permalink to KF2

KF3 CHG combined with BMI showed a stronger quartile association with incident CLMM than CHG alone within this analysis.

Comparing the highest and lowest CHG-BMI quartiles within the same cohort and fully adjusted model; relative effect-size ranking does not establish a causal role or clinical utility.

Evidence for KF3
Measure	Value
CHG-BMI Q4 versus Q1 adjusted OR	3.13; 95% CI 2.21–4.43
Comparison	CHG alone Q4 versus Q1: OR 1.64; 95% CI 1.22–2.22

Source locator: Results 3.2 (jdi70462-sec-0014); Table 2 (jdi70462-tbl-0002), CHG-BMI Q4 and CHG Q4, Model 2

Permalink to KF3

KF4 CHG-BMI had the largest observed AUC among the evaluated indices, with modest apparent discrimination.

Each index was evaluated in the same analytic sample without internal or external validation; the reported AUCs do not establish transportable clinical prediction.

Evidence for KF4
Measure	Value
CHG-BMI AUC	0.643; 95% CI 0.616–0.669
CHG AUC	0.575; 95% CI 0.546–0.603
TyG AUC	0.558; 95% CI 0.529–0.587
TG/HDL AUC	0.534; 95% CI 0.506–0.562

Source locator: Methods: Statistical analysis (jdi70462-sec-0011); Results 3.4 (jdi70462-sec-0016); Figure 3; Table 3 (jdi70462-tbl-0003)

Permalink to KF4

KF5 Jointly high CHG and BMI were associated with higher follow-up CLMM odds, without statistically significant interaction.

High/low groups used sample-specific median cut-offs and a low-CHG/low-BMI reference; these cut-offs are not established clinical thresholds.

Evidence for KF5
Measure	Value
High CHG plus high BMI adjusted OR	2.59; 95% CI 1.94–3.51
Multiplicative interaction	OR 0.90; 95% CI 0.58–1.40; P=0.643

Source locator: Results 3.5 (jdi70462-sec-0017), Joint effects and interaction of CHG and BMI; Table S4A–B

Permalink to KF5

Study Design & Model Development

Study profile

Study Design: Prospective observational analysis of the 2011 and 2015 CHARLS waves; baseline-index exposures and binary outcome ascertainment at follow-up; logistic regression, restricted cubic splines, and apparent ROC evaluation.
Evidence Type: Longitudinal observational association and unvalidated, in-sample discrimination
Unique total: 6,534
Population: Community-dwelling adults aged at least 45 years in mainland China, free of baseline CLMM; persons with zero or one component condition could enter the analysis.
Primary Endpoint: Newly meeting at least two criteria for CVD, type 2 diabetes, and LAP-based suspected MASLD at the 2015 follow-up.
Secondary Endpoint: Quartile associations, nonlinear relationships, apparent discrimination, joint CHG/BMI effects, and exploratory reclassification, subgroup, and sensitivity analyses.
External validation: No

Cohort hierarchy

Cohort	Role	n	Relationship
CHARLS analytic cohort	Single analysis cohort; not partitioned into independent development or validation populations	6,534	Independent cohort

Data modalities

Baseline fasting blood biochemistry
Anthropometric obesity indices
Interview and clinical covariates
Follow-up questionnaire, laboratory, and surrogate liver-phenotype ascertainment

Model development

Analytical Framework: Multivariable logistic regression and restricted cubic splines; index-specific ROC analysis
Performance Scope: Apparent discrimination in the same analysis sample; no internal or external validation
Endpoint Timing: Exact onset dates were unavailable; ORs are not hazard ratios
Clinical Thresholds: Sample-derived median and Youden cut-offs are not established clinical decision thresholds

What This Study Adds

Compares CHG with obesity- and inflammation-derived variants for a prospectively ascertained, surrogate-defined multimorbidity endpoint.
Separates the strength of adjusted associations from the modest, apparent discrimination of the evaluated indices.
Reports joint CHG/BMI associations while distinguishing them from evidence of statistical interaction.

Evidence Scope

Supports

Associations between baseline indices and meeting study-defined CLMM criteria at follow-up in this CHARLS analysis.
CHG-BMI ranking highest among evaluated indices for apparent ROC discrimination in this sample.
Hypothesis generation and independent validation of accessible metabolic-marker combinations.

Does Not Establish

A causal effect of CHG, obesity, or inflammation on multimorbidity.
Imaging- or biopsy-confirmed MASLD in every participant.
Hazard ratios, exact onset timing, or a time-to-event prediction model.
Independent internal or external validation, clinical net benefit, treatment efficacy, or a validated clinical cut-off.

Limitations

Residual confounding and observational design prevent causal inference.
Exact component-disease onset dates were unavailable; logistic ORs cannot be interpreted as hazards or incidence rates.
The MASLD component used an LAP-based suspected phenotype and may be misclassified; metabolic components overlap conceptually with the exposure indices.
The ROC analyses used the same analysis sample without internal or external validation, so apparent AUCs may be optimistic.
Single baseline exposure measurements, substantial complete-data selection, medication changes, correlated comparisons, and exploratory subgroup analyses limit interpretation.
Generalizability beyond Chinese adults aged at least 45 years requires independent evaluation.

When This Study Is Useful to Cite

CU1: What prospective population evidence links CHG-derived indices to incident CLMM?
Observational CHARLS associations with a surrogate-defined follow-up outcome, without causal inference.
Evidence: KF1, KF2, KF3
CU2: How well did CHG-BMI discriminate follow-up CLMM in the analysis sample?
Modest apparent discrimination without internal or external validation.
Evidence: KF4

What This Study Should Not Be Cited to Claim

A causal effect of CHG, obesity, or inflammation on multimorbidity.
Imaging- or biopsy-confirmed MASLD in every participant.
Hazard ratios, exact onset timing, or a time-to-event prediction model.
Independent internal or external validation, clinical net benefit, treatment efficacy, or a validated clinical cut-off.

Evidence Matrix

Row ID	Component	Context	Finding	Evidence Level	Scope	Source Locator	Evidence refs
evidence-jdi70462-1	A prospective CHARLS analysis identified incident, surrogate-defined cardiovascular–liver–metabolic multimorbidity at follow-up.	Chinese community-dwelling adults without baseline CLMM; component conditions were evaluated at baseline and the follow-up wave, rather than using exact onset dates.	Analytic participants: 6,534 adults aged 45 years or older; Exposure and outcome waves: 2011 baseline and 2015 follow-up; Incident CLMM: 406 participants; 6.2%; Outcome definition: At least two of CVD, type 2 diabetes, and LAP-based suspected MASLD	Prospective observational cohort analysis	Chinese community-dwelling adults without baseline CLMM; component conditions were evaluated at baseline and the follow-up wave, rather than using exact onset dates.	Methods: Study population and Outcome assessment (jdi70462-sec-0007; jdi70462-sec-0009); Results 3.1 (jdi70462-sec-0013); Figure 1; Table 1	KF1
evidence-jdi70462-2	Higher baseline CHG was associated with higher odds of meeting the CLMM criteria at follow-up.	Highest versus lowest CHG quartile in the fully adjusted logistic model; these are odds ratios for a binary follow-up outcome, not hazard ratios or treatment effects.	CHG Q4 versus Q1 adjusted OR: 1.64; 95% CI 1.22–2.22; Adjusted covariates: Age, sex, education, residence, marital status, smoking, drinking, hypertension, and CKD	Prospective observational cohort analysis	Highest versus lowest CHG quartile in the fully adjusted logistic model; these are odds ratios for a binary follow-up outcome, not hazard ratios or treatment effects.	Results 3.2 (jdi70462-sec-0014); Table 2 (jdi70462-tbl-0002), CHG Q4, Model 2	KF2
evidence-jdi70462-3	CHG combined with BMI showed a stronger quartile association with incident CLMM than CHG alone within this analysis.	Comparing the highest and lowest CHG-BMI quartiles within the same cohort and fully adjusted model; relative effect-size ranking does not establish a causal role or clinical utility.	CHG-BMI Q4 versus Q1 adjusted OR: 3.13; 95% CI 2.21–4.43; Comparison: CHG alone Q4 versus Q1: OR 1.64; 95% CI 1.22–2.22	Prospective observational cohort analysis	Comparing the highest and lowest CHG-BMI quartiles within the same cohort and fully adjusted model; relative effect-size ranking does not establish a causal role or clinical utility.	Results 3.2 (jdi70462-sec-0014); Table 2 (jdi70462-tbl-0002), CHG-BMI Q4 and CHG Q4, Model 2	KF3
evidence-jdi70462-4	CHG-BMI had the largest observed AUC among the evaluated indices, with modest apparent discrimination.	Each index was evaluated in the same analytic sample without internal or external validation; the reported AUCs do not establish transportable clinical prediction.	CHG-BMI AUC: 0.643; 95% CI 0.616–0.669; CHG AUC: 0.575; 95% CI 0.546–0.603; TyG AUC: 0.558; 95% CI 0.529–0.587; TG/HDL AUC: 0.534; 95% CI 0.506–0.562	Prospective observational cohort analysis	Each index was evaluated in the same analytic sample without internal or external validation; the reported AUCs do not establish transportable clinical prediction.	Methods: Statistical analysis (jdi70462-sec-0011); Results 3.4 (jdi70462-sec-0016); Figure 3; Table 3 (jdi70462-tbl-0003)	KF4
evidence-jdi70462-5	Jointly high CHG and BMI were associated with higher follow-up CLMM odds, without statistically significant interaction.	High/low groups used sample-specific median cut-offs and a low-CHG/low-BMI reference; these cut-offs are not established clinical thresholds.	High CHG plus high BMI adjusted OR: 2.59; 95% CI 1.94–3.51; Multiplicative interaction: OR 0.90; 95% CI 0.58–1.40; P=0.643	Prospective observational cohort analysis	High/low groups used sample-specific median cut-offs and a low-CHG/low-BMI reference; these cut-offs are not established clinical thresholds.	Results 3.5 (jdi70462-sec-0017), Joint effects and interaction of CHG and BMI; Table S4A–B	KF5

Q&A

Who was included in the analysis?

The final analytic cohort comprised 6,534 CHARLS participants aged at least 45 years without baseline CLMM. Baseline measurements were from 2011 and outcome ascertainment from 2015.

Evidence: KF1

What counted as cardiovascular–liver–metabolic multimorbidity?

Participants had to meet at least two criteria for CVD, type 2 diabetes, and LAP-based suspected MASLD at follow-up. The liver component was a surrogate phenotype, not universal imaging or histological confirmation.

Evidence: KF1

How strong was the association with CHG?

In the fully adjusted logistic model, the highest versus lowest CHG quartile had an OR of 1.64 with a 95% CI of 1.22–2.22. This is an observational odds ratio for the follow-up outcome.

Evidence: KF2

Did CHG-BMI perform better than CHG alone?

CHG-BMI had a stronger highest-versus-lowest-quartile association and the largest observed AUC among the compared indices. Its AUC of 0.643 was still modest and was evaluated in the analysis sample.

Evidence: KF3, KF4

Was this an externally validated clinical prediction model?

No. The article explicitly describes apparent, in-sample ROC evaluation without internal or external validation. Independent population validation and assessment beyond BMI and conventional risk factors are still needed.

Evidence: KF4

Did high CHG and high BMI interact statistically?

The jointly high group had greater follow-up CLMM odds than the low–low group, but the multiplicative interaction was not statistically significant. Joint association therefore does not establish synergy.

Evidence: KF5

Can these findings guide a treatment intended to prevent CLMM?

The study did not test an intervention. Observational associations and apparent discrimination do not establish that changing an index prevents disease or improves clinical decisions.

Evidence: KF2, KF3, KF4

Concepts & Entities

Conditions

Cardiovascular–liver–metabolic multimorbidityCardiovascular diseaseType 2 diabetesLAP-based suspected MASLD

Markers

Cholesterol–HDL–glucose indexCHG-BMITriglyceride–glucose index

Methods

CHARLSProspective observational cohortLogistic regressionApparent ROC discrimination

Related Research

AI hybrid survival assessment for advanced heart failure patients with renal dysfunction
A separate heart-failure prognosis study provides a contrast in endpoint and modeling purpose; it does not validate CHG-BMI or the CLMM endpoint. · DOI: 10.1038/s41467-024-50415-9
Plasma Olink Proteomics Reveals Novel Biomarkers for Prediction and Diagnosis in Dilated Cardiomyopathy with Heart Failure
A separate cardiac biomarker study uses a different disease population and measurement platform; it is related biomarker research rather than replication of this analysis. · DOI: 10.1021/acs.jproteome.4c00522

Publication & Provenance

DOI: https://doi.org/10.1111/jdi.70462
Publisher: https://onlinelibrary.wiley.com/doi/10.1111/jdi.70462
Doi Url: https://doi.org/10.1111/jdi.70462
PubMed: https://pubmed.ncbi.nlm.nih.gov/42852842/
PMCID: PMC13651581
Version Of Record: Journal of Diabetes Investigation; Early View, first published 9 October 2026
License: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)
License Url: https://creativecommons.org/licenses/by-nc/4.0/
Data Source: China Health and Retirement Longitudinal Study, 2011 baseline and 2015 follow-up
Supporting Material: Tables S1–S8 are linked in the version-of-record supplementary material; key estimates on this page are located in the full-text Results and principal tables.
Author Identity: Ge Zhang is listed with the Department of Cardiology, The First Affiliated Hospital of Zhengzhou University. The DOI is present in public ORCID 0000-0002-3116-3246; the self-maintained ORCID linkage is distinct from publisher-authenticated identity.
Evidence Basis: Version-of-record full text via Europe PMC XML; Methods, Results, Tables 1–3 and article-level permissions; PubMed and public ORCID metadata.

This is an original evidence synthesis linked to the published article. Cite the original article for study findings. The official abstract, when displayed separately, is reproduced verbatim under its article-level license; the synthesis preserves observational, endpoint-ascertainment, and validation limits.

Study-level evidence table

Study-level author-maintained evidence summary. Not participant-level data and not a ready-to-pool meta-analysis dataset. For formal citation, use the original article DOI.

Download study-level evidence CSV (UTF-8)

Formula safety: cells starting with =, +, -, @ (after leading whitespace), or tab/CR/LF are prefixed with an apostrophe. Treat all cells as text; remove only that safety prefix when recovering the original string.

All Publications Homepage