The index in question, known as the CHG index, is calculated as the natural logarithm of total cholesterol multiplied by fasting plasma glucose, divided by twice the HDL cholesterol level. Each component captures a different facet of metabolic health. Fasting glucose reflects how well the liver regulates blood sugar and how effectively pancreatic beta cells are functioning. Total cholesterol and HDL cholesterol together describe the lipid environment: high total cholesterol paired with low HDL is the classic signature of atherogenic dyslipidemia, the lipid disturbance that typically accompanies insulin resistance. By multiplying these values together, the index integrates lipid burden, reverse cholesterol transport capacity, and glycemic status into a single number that can be derived from blood tests already performed in virtually every health checkup.

What makes the new research notable is the population it examined. Most studies of metabolic risk indices have included people across all body mass index categories or focused on those who are overweight or obese. Lean individuals have been largely marginalized in metabolic research, precisely because their normal weight is assumed to signal metabolic health. Yet clinicians have long recognized a phenotype called metabolically obese normal weight—people who appear lean on the outside but harbor visceral fat accumulation and peripheral insulin resistance on the inside. Their cardiovascular and atherosclerotic risks can rival those of people with overt obesity, but because conventional risk scoring relies heavily on anthropometric measures like BMI, this population is chronically underidentified.

The research team, led by Junling Ding and Zhimao Cai of Shenzhen Second People’s Hospital, drew on a multicenter retrospective cohort of health examination participants screened between 2010 and 2016 at 32 sites across 11 Chinese cities. From an initial pool of more than 211,000 adults, they applied strict eligibility criteria: participants had to be at least 45 years old, have a BMI between 18.5 and 23.9 kilograms per square meter, and be free of prediabetes and diabetes at baseline. After excluding those with missing or implausible laboratory values, the final analytic cohort comprised 19,981 adults with a mean age of about 56 years, roughly 43 percent of them men. Participants were followed for an average of 3.19 years, during which the researchers tracked new cases of prediabetes, defined as fasting glucose reaching 5.6 millimoles per liter or higher, and self-reported diagnosed diabetes.

The results were unambiguous. In fully adjusted Cox proportional hazards models that accounted for age, sex, blood pressure, BMI, LDL cholesterol, kidney and liver markers, and family history of diabetes, each one-unit increase in the CHG index was associated with a 2.20-fold increase in the hazard of developing prediabetes or diabetes. When participants were divided into quartiles, those in the second, third, and fourth quartiles faced 42 percent, 86 percent, and 91 percent higher risks, respectively, compared with the lowest quartile. Kaplan-Meier survival curves showed the groups diverging early in follow-up, with the highest-quartile group consistently showing the lowest probability of remaining free of dysglycemia.

The most intriguing finding emerged when the researchers used smooth curve fitting and two-piecewise regression to probe the shape of the relationship. Rather than a straight line, the association displayed a clear inflection point at a CHG index value of 5.07. Below that threshold, each unit increase in the index was associated with a staggering 6.39-fold increase in risk. Above it, the association essentially vanished, with a hazard ratio of 1.02 that was statistically indistinguishable from no effect. A log-likelihood ratio test confirmed that the nonlinear model fit the data significantly better than a linear one.

The authors interpret this threshold effect as a possible biological saturation phenomenon. In the early phase of metabolic deterioration, rising fasting glucose, increasing total cholesterol, and falling HDL cholesterol jointly reflect accelerating insulin resistance and progressive beta-cell exhaustion, so each increment in the index corresponds to a disproportionately rapid rise in disease risk. Once metabolic dysregulation passes a critical severity, the system may have already entered a state of advanced decompensation, where further increases in the index no longer translate into additional risk elevation. In other words, the index may be most useful precisely in the window when intervention can still prevent disease—a potentially valuable property for a screening tool.

Subgroup analyses added further nuance. The association was stronger in women than in men, with hazard ratios of 2.65 and 1.77 respectively, and stronger in participants under 60 than in those 60 and older, with hazard ratios of 2.51 and 1.65. The authors suggest that sex differences in body fat distribution, which are well documented and linked to differential cardiometabolic risk, may explain the gender gap, while age-related changes such as sarcopenia, hormonal shifts, and cumulative exposure to multiple risk factors may dilute the relative contribution of the CHG index in older adults. Family history of diabetes did not meaningfully modify the association. A sensitivity analysis using E-values, which quantify how strong an unmeasured confounder would need to be to explain away an observed association, yielded a value of 3.82, suggesting the findings are reasonably robust.

Mechanistically, the index makes physiological sense. Elevated fasting glucose directly signals impaired hepatic glucose homeostasis and declining beta-cell function. In insulin-resistant states, increased hepatic de novo lipogenesis and very-low-density lipoprotein secretion, combined with reduced reverse cholesterol transport, produce exactly the pattern of raised total cholesterol and reduced HDL that pushes the index upward. These lipid abnormalities not only promote atherosclerosis but also exacerbate pancreatic lipotoxicity, further compromising beta-cell function. Chronic low-grade inflammation and oxidative stress, which are intertwined with both dyslipidemia and hyperglycemia, may mediate part of the pathway, and the authors note that related molecular mechanisms—including sirtuin proteins, inflammatory cascades, and programmed cell death pathways—are increasingly implicated in diabetic complications.

The study has limitations that temper enthusiasm. As a retrospective observational cohort, it cannot establish causality, and because the population was exclusively Chinese, generalizability to other ethnic groups remains uncertain. Residual confounding from unmeasured factors such as diet, physical activity, and socioeconomic status cannot be excluded, and smoking and drinking data were missing for more than 76 percent of participants, preventing their inclusion in adjusted models. The average follow-up of just over three years may also miss longer-term risk trajectories. Still, the core message is compelling: a number that costs almost nothing to compute from routine blood work may reveal hidden metabolic vulnerability in lean adults whom conventional screening would overlook. The authors call for large prospective cohorts to confirm the findings and establish causality, but if the results hold, the CHG index could become a practical tool for catching prediabetes early in a population that has too often been told its normal weight means it has nothing to worry about.

Subject of Research: The nonlinear association between the CHG metabolic index and incident prediabetes and diabetes in normal-weight middle-aged and older Chinese adults

Article Title: Nonlinear association of cholesterol, high-density lipoprotein, and glucose index with the risk of prediabetes and diabetes in normal-weight middle-aged and older Chinese adults: a multicenter retrospective cohort study

Article References: Ding, J., Cai, Z., Wang, X., Qin, L., & Chen, W. (2026). Nonlinear association of cholesterol, high-density lipoprotein, and glucose index with the risk of prediabetes and diabetes in normal-weight middle-aged and older Chinese adults: a multicenter retrospective cohort study. Frontiers in Nutrition, 13, Article 1949954. https://doi.org/10.3389/fnut.2026.1949954

Image Credits: AI Generated

DOI: 10.3389/fnut.2026.1949954

Keywords: CHG index, prediabetes, diabetes, normal weight, insulin resistance, HDL cholesterol, fasting glucose, cohort study, nonlinear association, threshold effect, metabolic health, China

The research team analyzed data from 1,379 adults with type 2 diabetes recruited in a hospital-based setting. Diabetic kidney disease was defined using standard clinical criteria: an estimated glomerular filtration rate below 60 mL/min/1.73 m², the presence of albuminuria with a urinary albumin-to-creatinine ratio of at least 30 mg/g, or both. Of the participants, 554 individuals, or 40.2 percent, met the definition of diabetic kidney disease. That high prevalence underscores why clinicians are keen to find markers that can be computed from routine laboratory panels rather than relying solely on more specialized or invasive assessments.

Both indices under investigation are built from everyday blood tests. The TyG index, which has gained considerable traction in metabolic research over the past decade, combines fasting triglycerides and fasting glucose into a logarithmic surrogate for insulin resistance. The CHG index is a newer proposal that blends total cholesterol, high-density lipoprotein cholesterol, and fasting glucose, aiming to capture a joint lipid-glycemic burden in a single number. The logic is that kidney damage in diabetes is driven not only by high blood sugar but also by lipid abnormalities and the metabolic stress they jointly impose on the delicate filtration barriers of the nephron.

To test whether CHG carries real signal, the researchers used multivariable logistic regression with progressive layers of covariate adjustment. In the primary model, which accounted for age, sex, body mass index, smoking, alcohol drinking, and diabetes duration, each one-standard-deviation increase in the CHG index was associated with 46 percent higher odds of diabetic kidney disease (odds ratio 1.46, 95 percent confidence interval 1.29 to 1.65). The TyG index performed almost identically, with an odds ratio of 1.44 per standard deviation (95 percent confidence interval 1.27 to 1.63). Both associations were statistically significant at P less than 0.001.

Crucially, the associations survived tougher scrutiny. When the team added systolic and diastolic blood pressure and the use of lipid-lowering medication to the model, and then further adjusted for glycated hemoglobin, the links between both indices and kidney disease remained robust. This matters because blood pressure and long-term glucose control are themselves major drivers of diabetic kidney damage; a marker that stays significant after accounting for them is capturing something beyond the usual suspects. Restricted cubic spline analysis, a flexible technique for mapping the shape of a dose-response relationship, showed significant overall associations for both indices without meaningful departures from linearity, with P values for nonlinearity of 0.348 for CHG and 0.904 for TyG. In other words, risk climbed steadily with each index rather than spiking only above some threshold.

The authors then went beyond simple association and asked how well each index actually discriminates between patients with and without diabetic kidney disease. Receiver operating characteristic analysis produced nearly identical areas under the curve: 0.684 for CHG and 0.683 for TyG, with a DeLong test P value of 0.715 confirming no statistically meaningful difference. An AUC around 0.68 indicates moderate discrimination, better than a coin flip but not accurate enough to serve as a stand-alone diagnostic test. The real question for any new index is whether it adds value on top of information clinicians already collect.

On that front, the exploratory results were cautiously encouraging. Adding CHG to a base clinical model yielded small but measurable improvements in reclassification, with an integrated discrimination improvement of 0.026 (95 percent confidence interval 0.012 to 0.046) and a continuous net reclassification improvement of 0.277 (95 percent confidence interval 0.160 to 0.392). Decision curve analysis, a method that evaluates the net benefit of a model across a range of decision thresholds, showed higher net benefit for the CHG-augmented model across threshold probabilities of 0.30 to 0.60. The authors were careful to flag that these were nominal P values not adjusted for multiple testing, so the reclassification findings should be treated as hypothesis-generating rather than definitive.

One of the more intriguing signals emerged from the prespecified subgroup analyses. A nominally significant interaction was observed for diabetes duration (P equal to 0.006), with the CHG–kidney disease association appearing stronger in patients who had lived with diabetes for ten years or longer. This pattern is biologically plausible: the microvascular damage that culminates in diabetic nephropathy typically accumulates over years of metabolic exposure, so a composite lipid-glycemic marker might exert a more visible statistical footprint in a longer-duration population. The analysis also indicated that the overall association was driven predominantly by albuminuria rather than by reduced glomerular filtration rate, pointing toward early injury of the glomerular filtration barrier as the relevant pathological process.

The study’s strengths lie in its reasonably large sample, its systematic comparison of two competing indices, and its use of modern analytical methods ranging from spline modeling to decision curve analysis. The authors also conducted exploratory sensitivity analyses and handled missing data with both multiple imputation and complete-case approaches to probe robustness. Still, the limitations are substantial and the authors state them plainly. The cross-sectional design means the data capture a single moment in time; elevated CHG and TyG are associated with prevalent kidney disease, but the study cannot establish that the indices predict who will develop it. Reverse causation remains possible, since declining kidney function itself alters lipid metabolism and glucose handling, potentially inflating the indices in people who already have kidney damage.

For now, the practical takeaway is that a marker computable from three routine blood values performs comparably to the TyG index in identifying patients with type 2 diabetes who have diabetic kidney disease, and may modestly sharpen risk assessment when layered onto standard clinical variables. The CHG index does not outshine its established rival, but matching it with a different lipid backbone gives researchers and clinicians a second, independent lens on the same metabolic territory. Whether that lens proves genuinely predictive will require prospective cohort studies that follow patients forward in time, ideally across multiple centers and ethnic groups. Until such validation arrives, the study stands as a careful, methodologically transparent demonstration of association, and a reminder that the intersection of blood lipids and blood sugar continues to hold clues to one of diabetes most consequential complications.

Subject of Research: Association of lipid-glycemic composite indices with diabetic kidney disease in type 2 diabetes

Article Title: Association of the CHG index with diabetic kidney disease and its discriminative performance in patients with type 2 diabetes: a cross-sectional comparison with the TyG index

Article References: Wang, T., Zhao, X., & Shi, R. (2026). Association of the CHG index with diabetic kidney disease and its discriminative performance in patients with type 2 diabetes: a cross-sectional comparison with the TyG index. BMC Endocrine Disorders. https://doi.org/10.1186/s12902-026-02620-4

Image Credits: AI Generated

DOI: 10.1186/s12902-026-02620-4

Keywords: diabetic kidney disease, type 2 diabetes, CHG index, TyG index, insulin resistance, albuminuria, cross-sectional study, logistic regression, ROC analysis, lipid metabolism, chronic kidney disease, biomarkers