Cross-sectional study links eGDR to cerebrovascular burden in non-diabetic adults

Lower estimated glucose disposal rate, a marker of insulin resistance derived from routine clinical measurements, was associated with a broader burden of subclinical cerebrovascular atherosclerosis in a cross-sectional study of 604 Chinese adults without diabetes. The researchers found that people with lower scores were more likely to have plaque or arterial stenosis affecting both intracranial and extracranial arterial beds, even though none had experienced a symptomatic cerebrovascular event. The association followed a linear pattern and persisted after adjustment for demographic, lifestyle and metabolic factors, but the study cannot establish whether insulin resistance precedes the vascular lesions or whether one causes the other.

The analysis used baseline data from the E-CHAMP prospective registry, which consecutively recruited community-dwelling residents from eight communities in Chengdu’s Xindu District between 2024 and 2026. Eligible participants were at least 45 years old, had no prior or current diagnosis of diabetes and had complete clinical, laboratory, anthropometric and imaging information. People with symptomatic cerebrovascular disease, including ischemic stroke or transient ischemic attack attributed to cerebrovascular stenosis, were excluded, as were individuals with coronary artery disease, severe liver or kidney dysfunction, psychiatric disorders or missing imaging and measurements required for the analysis. Of 800 enrolled adults, 604 were included; the mean age was 64 years, with a standard deviation of 11, and 220 participants, or 36.4%, were male.

Estimated glucose disposal rate, abbreviated eGDR, is a non-invasive composite surrogate for insulin resistance. The score combines hypertension status, waist-to-hip ratio and glycated hemoglobin, or HbA1c, to summarize metabolic conditions associated with the body’s ability to handle glucose. Higher eGDR values indicate better insulin sensitivity, whereas lower values indicate greater insulin resistance. The measure is especially relevant in people whose fasting glucose and HbA1c have not reached diabetic ranges but who may nevertheless harbor metabolic abnormalities associated with vascular injury. The investigators did not directly measure insulin resistance with a hyperinsulinemic-euglycemic clamp, so their results concern the composite metabolic status represented by eGDR rather than a directly quantified physiological state.

Cerebrovascular atherosclerosis was evaluated with carotid duplex ultrasonography, head-and-neck magnetic resonance angiography or head-and-neck computed tomography angiography, according to the imaging protocol and clinical practice. Two experienced neurologists independently reviewed the studies without access to participants’ clinical and laboratory data and resolved differences by consensus. The intracranial bed included the intracranial internal carotid, middle cerebral, anterior cerebral, posterior cerebral, vertebral and basilar arteries. The extracranial bed included the common carotid, extracranial internal carotid, vertebral and subclavian arteries. Involvement was defined as plaque and/or stenosis in a bed. Participants were then placed in three categories: no atherosclerosis, single vascular bed involvement, or involvement in both beds.

Of the 604 analyzed adults, 383, or 63.4%, had no cerebrovascular atherosclerosis, 164, or 27.2%, had disease in one vascular bed, and 57, or 9.4%, had disease in both beds. Participants with more extensive involvement were older and more likely to have hypertension, while their baseline eGDR values were lower. Across eGDR quartiles, the clearest gradient appeared in the prevalence of vascular involvement. Single-bed disease occurred in 33.1% of those in the lowest eGDR quartile, compared with 19.2% in the highest quartile. Dual-bed involvement fell from 13.2% in the lowest quartile to 2.0% in the highest. These are cross-sectional prevalence differences, not evidence that a change in eGDR would reduce existing disease.

Ordinal logistic regression was used to test whether higher eGDR corresponded to a lower probability of occupying a higher burden category. The unadjusted odds ratio was 0.67 per one-standard-deviation increase in eGDR, with a 95% confidence interval of 0.56 to 0.79. Adjustment for age and sex produced a similar estimate of 0.73, with a 95% confidence interval of 0.61 to 0.87. In the fully adjusted model, which additionally accounted for smoking, alcohol consumption, alanine aminotransferase, aspartate aminotransferase, uric acid, triglycerides, total cholesterol, HDL cholesterol and LDL cholesterol, the odds ratio was 0.74, with a 95% confidence interval of 0.62 to 0.89 and p=0.001. When eGDR was divided into quartiles, participants in the highest quartile had lower odds of greater burden than those in the lowest quartile: the adjusted odds ratio was 0.42, with a 95% confidence interval of 0.24 to 0.72 and p=0.001. The trend across quartiles was significant at p=0.002.

Restricted cubic spline analysis indicated that the dose–response relationship was consistent with a linear pattern. The overall association between eGDR and cerebrovascular burden was significant at p=0.002, while the test for nonlinearity was not significant at p=0.322. The analysis therefore did not detect a threshold, plateau or acceleration in the relationship within the observed distribution. A separate binary analysis that combined single- and dual-bed involvement into a single category of any cerebrovascular atherosclerosis also found an inverse association: the fully adjusted odds ratio per one-standard-deviation increase in eGDR was 0.72, with a 95% confidence interval of 0.60 to 0.87 and p<0.001. These results support a graded cross-sectional association rather than a pattern appearing only among people with the most advanced imaging abnormalities.

Subgroup analyses revealed that the relationship was generally similar across age, sex, body mass index, smoking status and LDL cholesterol category, but hypertension status modified the association. Among participants without hypertension, each one-standard-deviation increase in eGDR was associated with lower odds of greater atherosclerotic burden, with an odds ratio of 0.44, a 95% confidence interval of 0.23 to 0.85 and p=0.014. No significant association was detected among participants with hypertension: the odds ratio was 1.04, with a 95% confidence interval of 0.62 to 1.74 and p=0.892. The interaction was significant at p=0.003. Because hypertension is itself a component of the standard eGDR equation, the researchers repeated the analysis with a modified score that omitted hypertension. The interaction remained significant at p=0.002, suggesting it could not be explained solely by mathematical coupling, although the finding does not establish a biological interaction.

Component analyses further showed that hypertension was associated with higher cerebrovascular burden, with an odds ratio of 1.80 in a hypertension-only model and 1.79 in a model including hypertension, waist-to-hip ratio and HbA1c; both had p=0.001. Waist-to-hip ratio and HbA1c were not significantly associated with burden after mutual adjustment. The findings therefore cannot separate insulin resistance itself from the hypertension and adiposity information embedded in eGDR. They also do not show that eGDR provides useful predictive information beyond those components. Mechanistic explanations involving endothelial dysfunction, inflammation, oxidative stress, disturbed blood flow and vascular remodeling remain plausible, but the cross-sectional data cannot test them.

Several limitations constrain the conclusions. The single-region sample consisted mainly of middle-aged and older adults in Southwest China, and the findings may not generalize to populations with different body composition, metabolic profiles, vascular risks or healthcare systems. Roughly one-quarter of the initially eligible registry participants were excluded because imaging or anthropometric data were missing; although included and excluded groups had broadly similar baseline characteristics, selection bias cannot be ruled out. The use of three imaging modalities may also have introduced variation in lesion detection, despite predefined criteria and blinded review by two neurologists. Residual confounding, measurement error and possible misclassification remain possible. Most importantly, the cross-sectional design cannot determine temporal order or causation, and eGDR was not validated against a direct insulin-resistance test in this cohort. Prospective studies in more diverse populations, with repeated metabolic and vascular measurements, are needed to determine whether the association persists over time and whether eGDR offers value beyond its individual components.

Subject of Research: Association between estimated glucose disposal rate and subclinical cerebrovascular atherosclerotic burden in adults without diabetes

Article Title: Association between estimated glucose disposal rate and subclinical cerebrovascular atherosclerotic burden in non-diabetic middle-aged and older adults

Article References: Guo, L., Zhao, Y., Qu, M., Li, M., Zhu, J., Zheng, S., & Zhang, J. (2026). Association between estimated glucose disposal rate and subclinical cerebrovascular atherosclerotic burden in non-diabetic middle-aged and older adults. Frontiers in Neurology, 17, Article 1946802. https://doi.org/10.3389/fneur.2026.1946802

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1946802

Keywords: eGDR, insulin resistance, cerebrovascular atherosclerosis, intracranial stenosis, carotid plaque, vascular imaging, cross-sectional study, non-diabetic adults, hypertension, metabolic health, Association, between