This study leveraged data from the UK Biobank (2006–2010), a large prospective cohort encompassing approximately 500,000 individuals aged 40–69 years across the United Kingdom. Fig. 1 illustrates a detailed flowchart of the study participants’ selection. From the initial 502,389 participants, we excluded those with AAA at baseline, as well as individuals with missing waist circumference (WC), hypertension (HT) condition, glycated hemoglobin, and missing follow-up data. After applying these exclusion criteria, the analytical cohort comprised 416,800 participants.

Fig. 1.

Flow diagram of participant selection. HbA1c, glycated hemoglobin.

eGDR was calculated using the validated formula: eGDR = 21.158 – (0.09 $\times$ WC) – (3.407 $\times$ HT) – (0.551$\times$ HbA1c) [WC = waist circumference (cm), HT = hypertension status (1 = present, 0 = absent), and HbA1c = glycated hemoglobin] [16]. Participant data retrieved from the UK Biobank included sociodemographic factors (age, sex, ethnicity, education, environment), lifestyle behaviours (smoking, alcohol consumption), and clinical conditions (diabetes, hypertension, coronary heart disease). Anthropometric evaluations included vertical stature, weight assessment, and waist perimeter measurements. Laboratory investigations comprised fasting plasma glucose determination, HbA1c quantification, and detailed serum lipid profiling (measuring total cholesterol fractions, high-density lipoproteins, low-density lipoproteins, and triglyceride levels). We calculated the body mass index (BMI) using mass-to-height ratios (kg/m²).

The primary endpoint was defined as incident AAA identified through International Classification of Diseases, Tenth Revision (ICD-10) codes I71.3–I71.4. Patient identification involved comprehensive data collection from three sources: death registries, hospital records, and primary care databases. AAA cases were ascertained through linked records from death registries, hospital admissions, and primary care databases. The earliest recorded diagnosis in any of these sources was considered the time of AAA onset.

For covariates with sparse missing data (10%), continuous variables were completed using arithmetic means, while categorical variables were supplemented with their most frequent values. Participants were stratified into quartiles according to baseline eGDR levels, with cut-off values determined by the 25th, 50th, and 75th percentiles of the population distribution. Analytical comparisons across these quartiles incorporated the Kruskal-Wallis test for non-parametric continuous data, standard ANOVA for normally distributed variables, and chi-square testing for categorical outcomes.

Kaplan-Meier survival curves were generated to assess cumulative AAA risk, with group differences evaluated using log-rank tests. Cox proportional hazard regression models were constructed to quantify the association between eGDR and AAA risk. The analysis followed a stepwise adjustment protocol: Model 1: adjusted for sex and age; Model 2: further adjusted for drinking and smoking status; and Model 3: additionally adjusted for ethnic, education, BMI, diabetes, coronary heart disease (CHD), fasting blood glucose (FBG), total cholesterol (TC), triglycerides (TG), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C). We conducted separate analyses for each subgroup population using Model 3. Restricted cubic splines (RCS) were implemented to characterize potential nonlinear relationships. Time-dependent ROC curves were generated, and area under the curve (AUC) estimates were calculated to assess eGDR’s discriminative capacity.

Three sensitivity analyses were conducted to assess result robustness: (1) exclusion of participants with prevalent diabetes; (2) exclusion of AAA cases within 1-year follow-up; and (3) exclusion of AAA cases within 3-year follow-up. Schoenfeld residuals served to evaluate the proportional hazards assumption. All analyses were performed using R statistical software (version 4.3.0, RStudio, Inc., Boston, Massachusetts, America), implementing a significance threshold of p 0.05 (two-sided).

Table 1 presents the baseline characteristics of the 416,800 participants stratified by eGDR quartiles, including demographic parameters, metabolic profiles, and comorbid conditions. The cohort had a median age of 58.0 years. Participants in lower eGDR quartiles (reflecting greater insulin resistance) demonstrated progressively higher median age. Of the 416,800 study participants, 191,022 (45.83%) were male. The proportion of male participants was inversely correlated with eGDR quartiles, with a higher prevalence in the lower quartiles (p 0.001). Participants in the lower eGDR quartiles exhibited higher glycated haemoglobin levels, increased BMI, larger waist circumference, and elevated fasting glucose and triglyceride levels. These participants also showed lower levels of HDL cholesterol. Additionally, this group exhibited a greater frequency of tobacco use and concurrent medical conditions, encompassing diabetes, elevated blood pressure, and cardiac disease. Statistical analysis (p 0.001) confirmed the significance of these differences.

Fig. 2 presents Kaplan-Meier survival curves constructed for eGDR quartiles, demonstrating significantly higher cumulative AAA incidence in the lowest versus the highest quartile. Over a median follow-up period of 13.6 years, 1881 incident AAA cases were ascertained. RCS analysis revealed a nonlinear association between eGDR and AAA risk (nonlinearity p $\le$ 0.05), with a threshold effect observed at an eGDR value of 7.78, showing risk reduction above eGDR 7.78 mg/kg/min but accelerated risk elevation below this threshold (Fig. 3).

Fig. 2.

Cumulative incidence rate of AAA in 416,800 individuals separated by the four categories of eGDR. AAA, abdominal aortic aneurysm.

Fig. 3.

Restricted Cubic Spline to evaluate the association between eGDR and AAA.

In categorical analyses using eGDR quartiles, progressively reduced AAA risks were observed across higher quartiles compared to Q1 (reference): the hazard ratio (HR) (95% CI) of AAA in eGDR quartiles 2 (Q2), quartiles 3 (Q3), and quartiles 4 (Q4) were 0.44 (0.39–0.49), 0.28 (0.25–0.32), and 0.08 (0.06–0.10), respectively (p for trend 0.001; Table 2). This inverse association remained significant after adjusting for smoking and alcohol use (Model 2, p 0.001 for trend) and additional covariates, including ethnicity, education, environment, BMI, diabetes, and glucose and lipid profiles (Model 3, p 0.001 for trend). In the fully adjusted model (Model 3), Q4 participants exhibited 54% lower AAA risk versus Q1 [HR 0.46 (95% CI: 0.35–0.62), p 0.001 for trend]. When eGDR was analyzed as a continuous variable, each unit increment of eGDR was associated with a 27% lower AAA risk [HR 0.73 (95% CI: 0.72–0.75), p 0.001] in the unadjusted model. After full covariate adjustment, each unit increase of eGDR remained associated with a 12% lower AAA risk [HR 0.88 (95% CI: 0.85–0.90), p 0.001]. Sensitivity analyses demonstrated robustness of results (Table 2). When excluding patients diagnosed with diabetes, per unit increment of eGDR was associated with 12% lower AAA risk [HR 0.88 (95% CI: 0.84–0.91), p 0.001]. When excluding the population with short follow-up time, the eGDR on the risk of AAA was similar.

Fig. 4 presents stratified analyses evaluating the consistency of eGDR-AAA associations across demographic and clinical subgroups. The effect of eGDR on AAA risk was consistent across all subgroups. Consistent with known epidemiology, men exhibited a significantly higher incidence of AAA (0.84%) compared to women (0.13%). Sex-stratified analyses demonstrated comparable risk reduction per unit eGDR increase: Male: (HR 0.88, 95% CI: 0.85–0.92), Female: (HR 0.85, 95% CI: 0.78–0.92). In participants with hypertension, the multivariable-adjusted HR for AAA was 0.78 (95% CI: 0.72–0.86, p 0.001). Among those without hypertension, the corresponding HR was 0.85 (95% CI: 0.78–0.93, p 0.001). Subgroup analysis stratified by age groups revealed a significant interaction between age and the association of eGDR with AAA risk (p-interaction 0.001). Specifically, the effect of eGDR was more pronounced in participants aged 65 years (HR 0.81, 95% CI: 0.77–0.84) compared to those aged $\ge$65 years (HR 0.88, 95% CI: 0.84–0.93), indicating age-dependent heterogeneity in the magnitude of risk reduction. Subgroup analysis by HbA1c levels ($\ge$6.5% vs. 6.5%) showed consistent associations between eGDR and AAA. For HbA1c $\ge$6.5%, HR was 0.86 (95% CI: 0.76–0.97, p = 0.012); for HbA1c 6.5%, HR was 0.87 (95% CI: 0.84–0.90, p 0.001). No significant interaction was observed between HbA1c stratification and eGDR-AAA association (p-interaction = 0.835). Supplementary Table 1 further presents characteristics stratified by diabetes status, revealing distinct metabolic patterns between diabetic and non-diabetic subgroups. The total sample size in Supplementary Table 1 (N = 416,799) accounts for the exclusion of one participant with a missing baseline diagnosis value.

Fig. 4.

Subgroup analysis of eGDR on AAA.

As previously noted, the eGDR was developed as a composite parameter reflecting insulin resistance, comprising three clinical variables: waist circumference, hypertension status, and glycated hemoglobin. To investigate the relative contributions of these three components within the eGDR formula, we conducted standardized contribution assessment analyses, with results detailed in Supplementary Table 2. Based on the eGDR formula, HT (binary variable) had the largest absolute impact on eGDR values (–3.407 units if present). WC and HbA1c, as continuous variables, contributed less per unit change. In addition, the predictive performance of eGDR and its components for AAA risk was evaluated using receiver operating characteristic (ROC) analysis (Supplementary Fig. 1). The eGDR composite score demonstrated superior discriminative ability with an AUC of 0.715, significantly outperforming hypertension alone (AUC = 0.628, p 0.001), waist circumference (AUC = 0.693, p 0.001), and HbA1c (AUC = 0.632, p 0.001). These results statistically confirm that the eGDR provides incremental predictive value compared to hypertension, waist circumference, and HbA1c.

Although the hyperinsulinemic-euglycemic clamp remains the gold standard for assessing insulin resistance, its utility in large-scale epidemiological studies and routine clinical practice is constrained by procedural complexity. eGDR has been validated as a reliable marker of insulin resistance, particularly in Type 1 diabetes [6], and also has recently been recognised as a prognostic indicator for cardiovascular outcomes and mortality [11, 14]. Zheng et al. [16] demonstrated an inverse relationship between long-term eGDR levels and cardiovascular risk in both European and Asian populations. An analysis of individuals without diabetes mellitus indicated that lower levels of eGDR were associated with an increased risk of cardiovascular diseases among non-diabetic participants [14]. A prospective cohort study showed that a higher eGDR level was associated with a decreased risk of cardiovascular disease in non-diabetic chronic kidney disease patients [17]. A lower eGDR was associated with a higher risk of cardiovascular diseases in prediabetic populations [18], as was also found in the American population [19]. Yi et al. [20] demonstrated that each standard deviation increment in eGDR was associated with a 30% reduced likelihood of developing atherosclerotic cardiovascular disease (HR 0.70, 95% CI: 0.60–0.80). We demonstrated that higher eGDR values were inversely associated with AAA susceptibility, consistent with existing evidence on cardiovascular and atherosclerotic conditions. Restricted cubic spline analysis revealed a non-linear relationship between eGDR and AAA occurrence, mirroring patterns reported in previous cardiovascular studies [16]. These findings provide novel evidence linking eGDR to AAA susceptibility, positioning eGDR as a potential marker for both insulin resistance and AAA risk prediction.

Subgroup analysis indicated that eGDR may serve as a reliable biomarker for AAA risk prediction. The higher AAA incidence in males likely reflects sex-specific biological factors (e.g., hormonal profiles, abdominal wall anatomy), yet these factors did not modify the eGDR-AAA association. Further studies could explore sex-specific pathways linking insulin resistance to vascular pathogenesis. Hypertension-stratified analyses demonstrated consistent eGDR-AAA associations across subgroups, with numerically stronger effect sizes in hypertensive individuals. The observed age-dependent heterogeneity in eGDR-AAA associations highlights age-related modifications in metabolic-vascular interactions. In younger populations, metabolic plasticity may amplify insulin sensitivity benefits, whereas vascular aging, comorbidities, and polypharmacy in older adults may attenuate these effects. Notably, consistent eGDR effects were observed across age-based AAA screening criteria, reinforcing its predictive validity. eGDR could serve as an adjunct to ultrasound screening for high-risk individuals outside standard screening age ranges. For individuals with low eGDR (indicating high insulin resistance), abdominal ultrasound or contrast-enhanced computed tomography (CT) screening may be prioritized. HbA1c-stratified analyses revealed robust eGDR-AAA associations, excluding confounding by glycemic status and enhancing result credibility.

Emerging evidence indicates that impaired insulin sensitivity plays a pivotal role in the development of atherosclerotic vascular diseases [21]. Insulin resistance disrupts glucose metabolism, inducing oxidative stress, inflammation, and vascular cell apoptosis that collectively impair endothelial and smooth muscle function [22, 23]. Beyond glucose metabolism dysregulation, impaired insulin signaling disrupts lipid homeostasis, elevating blood lipid levels that potentiate both early and advanced atherosclerotic plaque formation [24, 25]. Given the established role of atherosclerosis in AAA pathogenesis [26], the eGDR-AAA association may be mediated through these metabolic pathways. Although insulin resistance damages vascular systems through oxidative stress, inflammation, cellular apoptosis, and lipid metabolism dysregulation, direct experimental evidence linking insulin resistance to AAA remains lacking. Future molecular and animal studies are required to explore this association, including developing animal models of insulin resistance or AAA, and quantifying inflammatory biomarkers in abdominal aortic tissues.

eGDR was developed as an effective indicator of insulin resistance [6]. Insulin resistance refers to a pathological state characterized by reduced target tissue responsiveness to circulating insulin [27]. Insulin resistance contributes to the pathogenesis of multiple disorders, including obesity [28], metabolic syndrome [29], non-alcoholic fatty liver disease [30], atherosclerosis [31], and diabetes mellitus (DM) [32]. Our findings demonstrate a dose-dependent relationship between insulin resistance severity and AAA risk. Sensitivity analyses excluding diabetic participants yielded consistent results. Paradoxically, it is known that diabetic patients tend to develop smaller AAA [33]. Based on the results of previous studies, the potential explanation was given as follows: (1) Insulin resistance represents a key pathophysiological feature of diabetes mellitus, preceding overt hyperglycemia [34]. The prediabetes-AAA association remains mechanistically undefined. Diabetes progression stages may exert differential effects on AAA pathogenesis, warranting deeper investigations. Future studies should examine eGDR-AAA associations across different glycemic control states. (2) Insulin resistance and hyperglycemia mediate vascular pathophysiology through distinct molecular pathways [35]. Thus, insulin resistance and diabetes mellitus differentially influence AAA pathogenesis. (3) A population-based cohort study demonstrated that oral hypoglycemic agents reduce AAA incidence in diabetics [36]. These agents ameliorate insulin resistance [37]. Pharmacological interventions may modulate the differential impacts of insulin resistance versus diabetes on AAA. This critical knowledge gap warrants prioritized investigation.