Estimated glucose disposal rate outperforms conventional insulin resistance indices in predicting acute kidney injury risk among individuals with diabetes and prediabetes Authors: Jinsheng Mai, Yiting Wu, Yiwei Zhang, Yanjun Zhang, Sisi Yang, Xianhui Qin, Yuanyuan Zhang, Jianping Jiang Published: 2026 ABSTRACT BackgroundAcute kidney injury (AKI) is a major global health concern, particularly in individuals with diabetes or prediabetes. While insulin resistance (IR) is implicated in AKI pathogenesis, the relationship between comprehensive IR measures—particularly the estimated glucose disposal rate (eGDR)—and AKI risk remains unclear.MethodsThis prospective cohort study analyzed 60,149 UK Biobank participants with diabetes or prediabetes. eGDR was calculated using waist circumference, hypertension status, and HbA1c. The primary outcome was incident AKI. Multivariable Cox models assessed the eGDR-AKI risk relationship. Predictive performance was compared against conventional IR indices using C-index, net reclassification improvement (NRI), and integrated discrimination improvement (IDI).ResultsOver a median follow-up duration of 13.4 years, 6,259 participants developed AKI. Higher eGDR showed a strong inverse association with AKI risk (per SD increment, adjusted HR: 0.64; 95%CI: 0.61–0.66), with a 62% lower risk in the highest vs. lowest quartile (adjusted HR: 0.38, 95%CI: 0.34–0.43). The eGDR showed superior discriminative ability (C-index 0.709, 95%CI: 0.702–0.715) compared to other IR indices - including the triglyceride-glucose index (TyG, 0.665, 0.658–0.672), TyG-waist circumference (0.693, 0.686-0.700), triglyceride-to-HDL cholesterol ratio (0.661, 0.654–0.668), C-reactive protein-triglyceride-glucose index (0.679, 0.672–0.686), single-point insulin sensitivity estimator index (SPISE, 0.680, 0.673–0.687), metabolic score for insulin resistance (METS-IR, 0.694, 0.687-0.700), and lipid accumulation product (LAP, 0.669, 0.662–0.676) - while also demonstrating clinically meaningful improvements in risk reclassification (NRI 0.063–0.196; IDI 0.003–0.018).ConclusionLower eGDR, indicating severe IR, independently predicts higher AKI risk in dysglycemic individuals. As a composite measure integrating metabolic and vascular components, eGDR outperforms conventional IR indices in AKI risk stratification, suggesting its clinical utility for identifying high-risk patients who may benefit from targeted interventions.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13098-026-02184-2. Introduction Acute kidney injury (AKI) poses a significant global health burden, with an estimated annual incidence of 13.3 million cases worldwide [1]. This sudden renal dysfunction precipitates life-threatening complications such as fluid overload, electrolyte disturbances, and systemic inflammation, while simultaneously elevating long-term risks of mortality and progression to chronic kidney disease (CKD) [2, 3]. Although renal replacement therapy has advanced considerably, the absence of effective pharmacological treatments underscores the critical need to identify novel, modifiable risk factors for AKI prevention. Diabetes mellitus (DM) is a well-established risk factor for AKI, with studies reporting a 2- to 8-fold higher incidence in diabetic individuals compared to non-diabetic populations [4, 5]. While hyperglycemia contributes directly to renal damage through oxidative stress and advanced glycation end-products, emerging evidence suggests that insulin resistance (IR) plays a central pathogenic role. IR disrupts renal microcirculation, impairs podocyte function, and promotes tubular injury via aberrant insulin signaling pathways [6]. Paradoxically, despite these mechanistic insights, clinical evidence linking IR to AKI risk remains scarce. To date, only one prospective study in type 2 diabetes patients undergoing coronary angiography has demonstrated an association between a high triglyceride-glucose (TyG) index and increased incidence of contrast-induced AKI (CI-AKI) [7]. Thus, the relationship between IR and AKI risk in broader diabetic and prediabetic populations remains poorly characterized. The estimated glucose disposal rate (eGDR) offers a novel approach to assessing IR. Unlike conventional surrogates, eGDR incorporates vascular (hypertension status, abdominal adiposity) and metabolic (glycated hemoglobin [HbA1c]) components, providing a more comprehensive reflection of IR’s multisystem pathophysiology. Accumulating evidence suggests that lower eGDR, indicating greater IR, is associated with cardiovascular events, microvascular complications, CKD progression, and mortality [8–14]. However, its relationship with AKI risk remains unexplored, and no studies have systematically compared its predictive performance against established IR indices—such as the triglyceride-to-high-density lipoprotein cholesterol ratio (TG/HDL), C-reactive protein-triglyceride-glucose index (CTI), single-point insulin sensitivity estimator index (SPISE), metabolic score for insulin resistance (METS-IR), lipid accumulation product (LAP) and TyG—in AKI risk stratification. To address these critical knowledge gaps, we leveraged the large-scale, prospective UK Biobank cohort to: (1) Investigate the dose-response relationship between eGDR and AKI risk; (2) Assess potential effect modifiers of this association; and (3) Compare eGDR’s predictive performance with conventional IR indices (TG/HDL, TyG, CTI, SPISE, METS-IR and LAP). Methods Data source and study population The UK Biobank is a population-based prospective cohort that enrolled approximately 500,000 participants aged 37–73 years from 22 assessment centers across England, Wales, and Scotland between 2006 and 2010 [15–19]. At baseline, participants completed detailed touch-screen questionnaires capturing sociodemographic characteristics, lifestyle factors, and medical history [15]. Standardized physical measurements were obtained, and biological samples were collected for laboratory analysis. The study received ethical approval from the North West Research Ethics Committee (06/MRE08/65), with all participants providing written informed consent. This study represents a secondary analysis of the prospective UK Biobank cohort. The hypothesis that eGDR is associated with AKI risk was formulated prior to data analysis; however, the specific analysis was not pre-registered. We have followed the STROBE guidelines for reporting observational studies. For this analysis, we identified 91,043 participants with diabetes or prediabetes at baseline. Diabetes was defined as either self-reported physician diagnosis [20] or HbA1c ≥ 6.5% (48 mmol/mol), while prediabetes was classified as HbA1c 5.7–6.4% (39–47 mmol/mol) [21]. We excluded participants with: (1) pre-existing AKI (n = 895), (2) incomplete data for eGDR calculation (hypertension status, waist circumference, or HbA1c; n = 2,937), or (3) missing covariate information (n = 27,062). The final analytical cohort comprised 60,149 eligible participants (Figure S1). Exposure assessment The estimated glucose disposal rate (eGDR, mg/kg/min) was calculated using the validated formula: eGDR = 21.158 - (0.09 × WC) - (3.407 × HT) - (0.551 × HbA1c), where WC denotes waist circumference (cm), HT represents hypertension status (1 = present; 0 = absent) [22]. Hypertension was defined by ICD-10 code I10, measured blood pressure (SBP ≥ 140 mmHg or DBP ≥ 90 mmHg), antihypertensive medication use, or self-reported diagnosis. HbA1c levels, determined via high-performance liquid chromatography, were converted from mmol/mol to percentage using the standard Eq. (0.09148 × HbA1c[mmol/mol] + 2.152) [23]. Participants were stratified into eGDR quartiles, with the lowest quartile (most severe insulin resistance) as the reference group. Ascertainment of AKI Incident AKI cases were identified through the International Classification of Diseases, 10th Revision (ICD-10) codes: I120, I139, K767, N055, N056, N059, N139, N170, N171, N172, N178, N179, N19, N990, P960, R392, T795 [24]. Case ascertainment utilized multiple data sources: primary care records, hospital admissions, death certificates, and self-reported medical history during follow-up visits. This diagnostic approach has been previously validated against KDIGO criteria in a UK population, demonstrating 95% positive predictive value [25]. Follow-up duration was calculated from baseline assessment until the first occurrence of: AKI diagnosis, death, loss to follow-up, or end of follow-up (October 31, 2022 for England; May 31, 2022 for Wales; August 31, 2022 for Scotland). Covariates measurement Comprehensive baseline data were collected through standardized questionnaires [15], physical examinations, and laboratory analyses, including demographic characteristics (age, sex, ethnicity), socioeconomic status (Townsend Deprivation Index, TDI), lifestyle factors (smoking status, alcohol consumption, physical activity), clinical parameters (body mass index [BMI], systolic blood pressure [SBP], diastolic blood pressure [DBP]), and medical history (hypertension, hypercholesterolemia, cardiovascular disease [CVD], glomerular or renal tubulo-interstitial diseases and CKD) [26–27]. Medication use was documented for antihypertensives, lipid-lowering agents, nonsteroidal anti-inflammatory drugs (NSAIDs) and hypoglycemic therapies. Laboratory measurements included lipid profile (triglycerides, total cholesterol, low-density lipoprotein cholesterol [LDL-C], HDL-C), glucose metabolism markers (glucose, HbA1c), inflammatory markers (C-reactive protein), albumin, and renal function parameters (serum creatinine, urinary albumin-to-creatinine ratio [UACR]). The eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation [28]. Baseline CKD was defined as self-reported CKD, pre-existing medical record of CKD, eGFR < 60 mL/min/1.73 m², or UACR ≥ 30 mg/g. Baseline glomerular or renal tubulo-interstitial diseases was identified through the ICD-10 codes: N00-N08 and N10-N16. Statistical analysis Baseline characteristics were presented as mean (standard deviation, SD) for continuous variables and as frequencies (percentages) for categorical variables. Differences across eGDR categories were assessed using ANOVA for continuous variables and chi-square (χ2) tests for categorical variables. We employed restricted cubic spline (RCS) Cox regression with 5 knots to characterize the nonlinear association between eGDR and AKI risk in the diabetes/prediabetes population. Multivariable Cox proportional hazards models were constructed to estimate adjusted hazard ratios (aHRs) and 95% confidence intervals (CIs), incorporating adjustments for: demographic factors (age, sex, ethnicity), socioeconomic and lifestyle variables (TDI, education, physical activity, smoking, alcohol consumption); and clinical parameters (BMI, C-reactive protein, eGFR, UACR) and comorbidities (CVD, CKD, hypercholesterolemia, glomerular or renal tubulo-interstitial diseases) and Medication use (NSAIDs). Cumulative AKI incidence was visualized using Kaplan-Meier curves with log-rank testing for trend across eGDR categories. Stratified analyses were performed to assess potential effect modification by key demographic and clinical factors, including age (< 65 vs. ≥65 years), sex (male vs. female), BMI (< 25 vs. 25–<30 vs. ≥30 kg/m²), smoking status (never vs. former vs. current), alcohol consumption (non-daily vs. daily), ethnicity (non-White vs. White), TDI ( < − 1.85 vs. ≥−1.85, median split), physical activity (low vs. moderate vs. high), education (non-university vs. university degree), hypertension (no vs. yes), hypercholesterolemia (no vs. yes), CVD (no vs. yes), CKD (no vs. yes), glomerular or renal tubulo-interstitial diseases (no vs. yes) and use of NSAIDs (no vs. yes). Statistical interaction was tested using multiplicative interaction terms. To compare the predictive performance of eGDR against alternative insulin resistance (IR) indices—including the TyG, TyG-WC, TG/HDL ratio, CTI, SPISE, METS-IR and LAP—we calculated the following metrics: (1) Harrell’s C-index for discrimination, (2) 10-year integrated discrimination improvement (IDI), and (3) 10-year continuous net reclassification improvement (NRI). The TyG index was derived as ln (triglycerides [mg/dl] × glucose [mg/dl]/2), while TyG-WC was computed as TyG × waist circumference (cm). The TG/HDL ratio was calculated as triglycerides (mg/dl) divided by HDL-C (mg/dl). The CTI index was determined using the formula: 0.412 × ln (CRP [mg/L]) + TyG [29]. The SPISE index was calculated using the formula: 600 × HDL-C (mg/dl)0.185/Triglycerides (mg/dl)0.2 × BMI (kg/m²)1.338 [30]. The METS-IR was defined as ln (2 × glucose [mg/dl]+triglycerides [mg/dl])×BMI (kg/m²) / ln(HDL-C [mg/dl]) [31]. The LAP index was calculated using the formula: (waist circumference [cm] − 65) × triglycerides (mmol/L) for men and (waist circumference [cm] − 58) × triglycerides (mmol/L) for women [32]. To ensure the robustness of the findings, we performed multiple sensitivity analyses: (1) excluding participants with pre-existing CVD or CKD at baseline; (2) removing AKI cases detected during the initial year, the first two years, the first three years, and the first five years of follow-up; (3) incorporating additional adjustments for serum biomarkers (including HDL-C, LDL-C, glucose, and albumin) and medication use (antihypertensive and lipid-lowering agents); and (4) redefining hypertension using updated criteria (systolic blood pressure ≥ 130 mmHg and/or diastolic blood pressure ≥ 80 mmHg) [33]. All statistical analyses were conducted using R software (version 4.4.2; R Foundation for Statistical Computing, available at http://www.R-project.org, accessed October 31, 2024). Statistical significance was defined as a two-tailed P-value < 0.05 for all tests. Results Study participants and baseline characteristics The study comprised 60,149 participants, including 31,686 (52.7%) males, with a mean age of 59.5 years (SD, 7.1). Compared to participants with the lowest eGDR, those with higher eGDR were more likely to be female, of non-white ethnicity, and non-smokers. They also had lower TDI scores, BMI, and blood pressure levels, along with higher educational attainment and physical activity levels. Furthermore, the high-eGDR group exhibited more favorable metabolic profiles, characterized by lower serum triglycerides, glucose, CRP, and UACR, as well as higher eGFR. This group also had a lower prevalence of comorbidities, including CVD, CKD, glomerular or renal tubulo-interstitial diseases and hypercholesterolemia, together with a lower use of NSAIDs (Table 1). Table 1Baseline Characteristics of Participants with Diabetes or Prediabetes by Estimated Glucose Disposal Rate (eGDR) Quartiles *Quartile of eGDR, mg/kg/minP valueQuartile 1(< 4.9)Quartile 2(4.9- <6.1)Quartile 3(6.1- <8.2)Quartile 4(≥ 8.2) N 15,03815,03715,03715,037Age, y59.6 (6.9)60.6 (6.7)60.4 (6.7)57.3 (7.5)< 0.001Male, n (%)10,809 (71.9)9445 (62.8)5808 (38.6)5624 (37.4)< 0.001White, n (%)13,914 (92.5)13,680 (91.0)13,568 (90.2)13,457 (89.5)< 0.001 Townsend deprivation index -0.6 (3.3)-1.0 (3.2)-1.3 (3.1)-1.2 (3.2)< 0.001College or University degree, n (%)3605 (24.0)3872 (25.7)4158 (27.7)5278 (35.1)< 0.001Body mass index, kg/m234.9 (5.3)29.8 (3.5)27.1 (4.2)26.3 (3.9)< 0.001Systolic blood pressure, mmHg145.9 (17.1)147.5 (17.1)146.0 (17.5)125.8 (10.7)< 0.001Diastolic blood pressure, mmHg85.7 (10.3)85.6 (10.0)83.8 (9.7)76.4 (7.1)< 0.001Physical activity, n (%)< 0.001Low4808 (32.0)3220 (21.4)2613 (17.4)2805 (18.7)Moderate5879 (39.1)6304 (41.9)6243 (41.5)6188 (41.2)High4351 (28.9)5513 (36.7)6181 (41.1)6044 (40.2)Smoking status, n (%)< 0.001Never6075 (40.4)6637 (44.1)7897 (52.5)7883 (52.4)Previous7196 (47.9)6493 (43.2)5089 (33.8)4527 (30.1)Current1767 (11.8)1907 (12.7)2051 (13.6)2627 (17.5) Daily or almost daily (%) 2538 (16.9)3065 (20.4)2767 (18.4)2433 (16.2)< 0.001 Laboratory indicators Triglyceride, mg/dl2.4 (1.3)2.2 (1.2)1.9 (1.0)1.8 (1.0)< 0.001Cholesterol, mg/dl4.9 (1.2)5.3 (1.3)5.7 (1.3)5.8 (1.2)< 0.001Low-density lipoprotein cholesterol, mg/dl3.1 (0.9)3.3 (1.0)3.5 (1.0)3.6 (0.9)< 0.001High-density lipoprotein cholesterol, mg/dl1.1 (0.3)1.3 (0.3)1.4 (0.4)1.4 (0.4)< 0.001C-reactive protein, mg/L4.5 (5.5)3.5 (5.0)3.1 (5.1)2.8 (5.1)< 0.001Albumin, g/L44.6 (2.7)45.2 (2.6)45.1 (2.7)44.8 (2.6)< 0.001Glucose, mmol/L7.2 (3.3)5.8 (1.9)5.6 (1.7)5.3 (1.5)< 0.001HbA1c, %6.9 (1.4)6.1 (0.6)6.0 (0.6)6.0 (0.4)< 0.001Urinary albumin creatinine ratio, mg/g36.4 (168.5)19.9 (98.5)17.1 (68.2)11.8 (34.7)< 0.001Estimated glomerular filtration rate, ml/min/1.73m288.9 (15.6)87.1 (14.2)88.1 (13.7)90.7 (12.9)< 0.001Disease history, n (%)Hypertension14,980 (99.6)14,852 (98.8)12,817 (85.2)446 (3.0)< 0.001Cardiovascular disease3352 (22.3)2661 (17.7)1765 (11.7)786 (5.2)< 0.001Chronic kidney disease2921 (19.4)1785 (11.9)1504 (10.0)797 (5.3)< 0.001Hypercholesterolemia4861 (32.3)4595 (30.6)3455 (23.0)1890 (12.6)< 0.001Glomerular or renal tubulo-interstitial diseases140 (0.9)98 (0.7)93 (0.6)56 (0.4)< 0.001Medication use, n (%)Nonsteroidal anti-inflammatory drugs9239 (61.4)7986 (53.1)6828 (45.4)5622 (37.4)< 0.001*Variables are presented as Mean (SD) or n (%) Table 1Baseline Characteristics of Participants with Diabetes or Prediabetes by Estimated Glucose Disposal Rate (eGDR) Quartiles *Quartile of eGDR, mg/kg/minP valueQuartile 1(< 4.9)Quartile 2(4.9- <6.1)Quartile 3(6.1- <8.2)Quartile 4(≥ 8.2) N 15,03815,03715,03715,037Age, y59.6 (6.9)60.6 (6.7)60.4 (6.7)57.3 (7.5)< 0.001Male, n (%)10,809 (71.9)9445 (62.8)5808 (38.6)5624 (37.4)< 0.001White, n (%)13,914 (92.5)13,680 (91.0)13,568 (90.2)13,457 (89.5)< 0.001 Townsend deprivation index -0.6 (3.3)-1.0 (3.2)-1.3 (3.1)-1.2 (3.2)< 0.001College or University degree, n (%)3605 (24.0)3872 (25.7)4158 (27.7)5278 (35.1)< 0.001Body mass index, kg/m234.9 (5.3)29.8 (3.5)27.1 (4.2)26.3 (3.9)< 0.001Systolic blood pressure, mmHg145.9 (17.1)147.5 (17.1)146.0 (17.5)125.8 (10.7)< 0.001Diastolic blood pressure, mmHg85.7 (10.3)85.6 (10.0)83.8 (9.7)76.4 (7.1)< 0.001Physical activity, n (%)< 0.001Low4808 (32.0)3220 (21.4)2613 (17.4)2805 (18.7)Moderate5879 (39.1)6304 (41.9)6243 (41.5)6188 (41.2)High4351 (28.9)5513 (36.7)6181 (41.1)6044 (40.2)Smoking status, n (%)< 0.001Never6075 (40.4)6637 (44.1)7897 (52.5)7883 (52.4)Previous7196 (47.9)6493 (43.2)5089 (33.8)4527 (30.1)Current1767 (11.8)1907 (12.7)2051 (13.6)2627 (17.5) Daily or almost daily (%) 2538 (16.9)3065 (20.4)2767 (18.4)2433 (16.2)< 0.001 Laboratory indicators Triglyceride, mg/dl2.4 (1.3)2.2 (1.2)1.9 (1.0)1.8 (1.0)< 0.001Cholesterol, mg/dl4.9 (1.2)5.3 (1.3)5.7 (1.3)5.8 (1.2)< 0.001Low-density lipoprotein cholesterol, mg/dl3.1 (0.9)3.3 (1.0)3.5 (1.0)3.6 (0.9)< 0.001High-density lipoprotein cholesterol, mg/dl1.1 (0.3)1.3 (0.3)1.4 (0.4)1.4 (0.4)< 0.001C-reactive protein, mg/L4.5 (5.5)3.5 (5.0)3.1 (5.1)2.8 (5.1)< 0.001Albumin, g/L44.6 (2.7)45.2 (2.6)45.1 (2.7)44.8 (2.6)< 0.001Glucose, mmol/L7.2 (3.3)5.8 (1.9)5.6 (1.7)5.3 (1.5)< 0.001HbA1c, %6.9 (1.4)6.1 (0.6)6.0 (0.6)6.0 (0.4)< 0.001Urinary albumin creatinine ratio, mg/g36.4 (168.5)19.9 (98.5)17.1 (68.2)11.8 (34.7)< 0.001Estimated glomerular filtration rate, ml/min/1.73m288.9 (15.6)87.1 (14.2)88.1 (13.7)90.7 (12.9)< 0.001Disease history, n (%)Hypertension14,980 (99.6)14,852 (98.8)12,817 (85.2)446 (3.0)< 0.001Cardiovascular disease3352 (22.3)2661 (17.7)1765 (11.7)786 (5.2)< 0.001Chronic kidney disease2921 (19.4)1785 (11.9)1504 (10.0)797 (5.3)< 0.001Hypercholesterolemia4861 (32.3)4595 (30.6)3455 (23.0)1890 (12.6)< 0.001Glomerular or renal tubulo-interstitial diseases140 (0.9)98 (0.7)93 (0.6)56 (0.4)< 0.001Medication use, n (%)Nonsteroidal anti-inflammatory drugs9239 (61.4)7986 (53.1)6828 (45.4)5622 (37.4)< 0.001*Variables are presented as Mean (SD) or n (%) Baseline Characteristics of Participants with Diabetes or Prediabetes by Estimated Glucose Disposal Rate (eGDR) Quartiles * *Variables are presented as Mean (SD) or n (%) Association between eGDR and incident AKI risk During a median follow-up period of 13.4 years, 6,259 participants (10.4%) developed incident AKI. The study revealed a significant inverse relationship between eGDR and AKI risk among individuals with diabetes or prediabetes, with each standard deviation increase in eGDR associated with a 36% lower risk of AKI (adjusted HR: 0.64; 95% CI: 0.61–0.66). Participants in the highest eGDR quartile demonstrated a substantially reduced AKI risk compared to those in the lowest quartile (adjusted HR: 0.38; 95% CI: 0.34–0.43) (Fig. 1; Table 2). Consistent results were also obtained in the diabetic or prediabetic population when analyzed separately (Table S1, Figure S2). Fig. 1Dose-Response Relationship Between Estimated Glucose Disposal Rate (eGDR) and Incident Acute Kidney Injury Risk in Diabetes and Prediabetes* *Adjusted for age, sex, ethnicity, Townsend Deprivation Index, body mass index, systolic blood pressure, diastolic blood pressure, education, physical activity, smoking status, alcohol consumption, C-reactive protein, estimated glomerular filtration rate, urinary albumin creatinine ratio, history of cardiovascular disease, chronic kidney disease, glomerular or renal tubulo-interstitial diseases and hypercholesterolemia, use of nonsteroidal anti-inflammatory drugs Dose-Response Relationship Between Estimated Glucose Disposal Rate (eGDR) and Incident Acute Kidney Injury Risk in Diabetes and Prediabetes* *Adjusted for age, sex, ethnicity, Townsend Deprivation Index, body mass index, systolic blood pressure, diastolic blood pressure, education, physical activity, smoking status, alcohol consumption, C-reactive protein, estimated glomerular filtration rate, urinary albumin creatinine ratio, history of cardiovascular disease, chronic kidney disease, glomerular or renal tubulo-interstitial diseases and hypercholesterolemia, use of nonsteroidal anti-inflammatory drugs Table 2Association Between Estimated Glucose Disposal Rate (eGDR) and Incident Acute Kidney Injury Risk in Participants with Diabetes or PrediabeteseGDR, mg/kg/min N Cases (%)Model 1 *Model 2 †HR (95% CI)P valueHR (95% CI)P value Per SD increment 60,1496259 (10.4)0.55(0.54,0.57)< 0.0010.64(0.61,0.66)< 0.001 Categories Quartile 1 (< 4.9)15,0382818 (18.7)RefRef Quartile 2 (4.9- <6.1)15,0371709 (11.4)0.54(0.51,0.58)< 0.0010.68(0.64,0.73)< 0.001 Quartile 3 (6.1- <8.2)15,0371095 (7.3)0.37(0.34,0.39)< 0.0010.53(0.49,0.58)< 0.001 Quartile 4 (≥ 8.2)15,037637 (4.2)0.25(0.23,0.27)< 0.0010.38(0.34,0.43)< 0.001P for trend< 0.001< 0.001* Model 1: adjusted for age, sex, ethnicity† Model 2: adjusted for age, sex, ethnicity, Townsend Deprivation Index, body mass index, systolic blood pressure, diastolic blood pressure, education, physical activity, smoking status, alcohol consumption, C-reactive protein, estimated glomerular filtration rate, urinary albumin creatinine ratio, history of cardiovascular disease, chronic kidney disease, glomerular or renal tubulo-interstitial diseases and hypercholesterolemia, use of nonsteroidal anti-inflammatory drugs Table 2Association Between Estimated Glucose Disposal Rate (eGDR) and Incident Acute Kidney Injury Risk in Participants with Diabetes or PrediabeteseGDR, mg/kg/min N Cases (%)Model 1 *Model 2 †HR (95% CI)P valueHR (95% CI)P value Per SD increment 60,1496259 (10.4)0.55(0.54,0.57)< 0.0010.64(0.61,0.66)< 0.001 Categories Quartile 1 (< 4.9)15,0382818 (18.7)RefRef Quartile 2 (4.9- <6.1)15,0371709 (11.4)0.54(0.51,0.58)< 0.0010.68(0.64,0.73)< 0.001 Quartile 3 (6.1- <8.2)15,0371095 (7.3)0.37(0.34,0.39)< 0.0010.53(0.49,0.58)< 0.001 Quartile 4 (≥ 8.2)15,037637 (4.2)0.25(0.23,0.27)< 0.0010.38(0.34,0.43)< 0.001P for trend< 0.001< 0.001* Model 1: adjusted for age, sex, ethnicity† Model 2: adjusted for age, sex, ethnicity, Townsend Deprivation Index, body mass index, systolic blood pressure, diastolic blood pressure, education, physical activity, smoking status, alcohol consumption, C-reactive protein, estimated glomerular filtration rate, urinary albumin creatinine ratio, history of cardiovascular disease, chronic kidney disease, glomerular or renal tubulo-interstitial diseases and hypercholesterolemia, use of nonsteroidal anti-inflammatory drugs Association Between Estimated Glucose Disposal Rate (eGDR) and Incident Acute Kidney Injury Risk in Participants with Diabetes or Prediabetes * Model 1: adjusted for age, sex, ethnicity † Model 2: adjusted for age, sex, ethnicity, Townsend Deprivation Index, body mass index, systolic blood pressure, diastolic blood pressure, education, physical activity, smoking status, alcohol consumption, C-reactive protein, estimated glomerular filtration rate, urinary albumin creatinine ratio, history of cardiovascular disease, chronic kidney disease, glomerular or renal tubulo-interstitial diseases and hypercholesterolemia, use of nonsteroidal anti-inflammatory drugs Kaplan-Meier analysis showed a graded association, with progressively higher cumulative AKI incidence observed across decreasing eGDR quartiles, particularly notable in the lowest quartile group (Fig. 2, Figure S3). Fig. 2Kaplan-Meier Curves of Acute Kidney Injury Incidence Stratified by Estimated Glucose Disposal Rate (eGDR) in Participants with Diabetes or Prediabetes Kaplan-Meier Curves of Acute Kidney Injury Incidence Stratified by Estimated Glucose Disposal Rate (eGDR) in Participants with Diabetes or Prediabetes Stratified analyses by potential effect modifiers Subgroup analyses were performed to assess the consistency of the eGDR-AKI association across different populations (Fig. 3). No significant effect modification was observed for age (< 65 vs. ≥65 years), alcohol consumption (non-daily vs. daily), TDI (median, < -1.85 vs. ≥ -1.85), hypercholesterolemia (no vs. yes), CKD (no vs. yes), glycemic status (prediabetes vs. diabetes) and NSAIDs (no vs. yes) (all P for interaction > 0.05; Fig. 3). Although statistically significant interactions (P < 0.05) were noted for sex(males vs. females), BMI (< 25 vs. 25- <30 vs. ≥30 kg/m²), smoking status (never vs. previous vs. current), ethnicity (non-white vs. white), physical activity (low vs. moderate vs. high), education (others vs. university degree), hypertension (no vs. yes), CVD (no vs. yes) and glomerular or renal tubulo-interstitial diseases (no vs. yes), the findings lacked clear clinical relevance due to consistent directional associations and multiple testing. Fig. 3Stratified Analysis: Association Between Estimated Glucose Disposal Rate (eGDR) and Incident Acute Kidney Injury Risk Across Subgroups in Diabetes and Prediabetes** Adjusted for age, sex, ethnicity, Townsend Deprivation Index, body mass index, systolic blood pressure, diastolic blood pressure, education, physical activity, smoking status, alcohol consumption, C-reactive protein, estimated glomerular filtration rate, urinary albumin creatinine ratio, history of cardiovascular disease, chronic kidney disease, glomerular or renal tubulo-interstitial diseases and hypercholesterolemia, use of nonsteroidal anti-inflammatory drugs Stratified Analysis: Association Between Estimated Glucose Disposal Rate (eGDR) and Incident Acute Kidney Injury Risk Across Subgroups in Diabetes and Prediabetes** Adjusted for age, sex, ethnicity, Townsend Deprivation Index, body mass index, systolic blood pressure, diastolic blood pressure, education, physical activity, smoking status, alcohol consumption, C-reactive protein, estimated glomerular filtration rate, urinary albumin creatinine ratio, history of cardiovascular disease, chronic kidney disease, glomerular or renal tubulo-interstitial diseases and hypercholesterolemia, use of nonsteroidal anti-inflammatory drugs Risk prediction performance for incident AKI The eGDR demonstrated superior predictive performance for AKI risk compared to conventional metabolic indices. It achieved a significantly higher discrimination accuracy (C-index, 0.709; 95%CI: 0.702, 0.715) than the TyG (C-index, 0.665; 95%CI: 0.658, 0.672; C-index increase, -0.044; 95%CI: -0.053, -0.034), TyG-WC (C-index, 0.693; 95%CI: 0.686, 0.6700; C-index increase, -0.016; 95%CI: -0.025, -0.006), TG/HDL (C-index, 0.661; 95%CI: 0.654, 0.668; C-index increase, -0.048; 95%CI: -0.057, -0.038), CTI (C-index, 0.679; 95%CI: 0.672, 0.686; C-index increase, -0.030; 95%CI: -0.039, -0.020), SPISE (C-index, 0.680; 95%CI: 0.673, 0.687; C-index increase, -0.029; 95%CI: -0.038, -0.019), METS-IR (C-index, 0.694; 95%CI: 0.687, 0.700; C-index increase, -0.015; 95%CI: -0.025, -0.006), and LAP (C-index, 0.669; 95%CI: 0.662, 0.676; C-index increase, -0.040; 95%CI: -0.049, -0.030) (Table 3). Furthermore, eGDR provided clinically meaningful improvement in risk stratification, as evidenced by NRI (0.063–0.196) and IDI (0.003–0.018) values when compared to these alternative indices (Table 3). Table 3Comparative Predictive Performance of Insulin Resistance Indices for Incident Acute Kidney Injury in Diabetes and PrediabetesModelC index (95%CI)C index increase (95%CI)10-y IDI improvement (95%CI)10-y continuous NRI (95% CI)Reference model: basic factors * + eGDR0.709(0.702–0.715) Basic factors + TyG0.665(0.658–0.672)-0.044(-0.053, -0.034)-0.016(-0.017, -0.003)-0.179(-0.197, -0.037) Basic factors + TyG-WC0.693(0.686-0.700)-0.016(-0.025, -0.006)-0.004(-0.006, -0.001)-0.077(-0.096, -0.014) Basic factos + TG/HDL0.661(0.654–0.668)-0.048(-0.057, -0.038)-0.018(-0.020, -0.004)-0.196(-0.213, -0.048) Basic factors + CTI0.679(0.672–0.686)-0.030(-0.039, -0.020)-0.011(-0.013, -0.002)-0.123(-0.149, -0.028) Basic factors + SPISE0.680(0.673–0.687)-0.029(-0.038, -0.019)-0.011(-0.013, -0.002)-0.136(-0.152, -0.026) Basic factors + METS-IR0.694(0.687-0.700)-0.015(-0.025, -0.006)-0.003(-0.005, 0.002)-0.063(-0.080, -0.005) Basic factors + LAP0.669(0.662–0.676)-0.040(-0.049, -0.030)-0.015(-0.018, -0.003)-0.189(-0.201, -0.044)* Basic factors including age, sex, ethnicity. Abbreviations: TyG, triglyceride-glucose index; TyG-WC, TyG- waist circumference; TG/HDL, triglyceride-to-HDL cholesterol ratio; CTI, C-reactive protein-triglyceride-glucose index; SPISE, single-point insulin sensitivity estimator index; METS-IR, metabolic score for insulin resistance; LAP, lipid accumulation product Table 3Comparative Predictive Performance of Insulin Resistance Indices for Incident Acute Kidney Injury in Diabetes and PrediabetesModelC index (95%CI)C index increase (95%CI)10-y IDI improvement (95%CI)10-y continuous NRI (95% CI)Reference model: basic factors * + eGDR0.709(0.702–0.715) Basic factors + TyG0.665(0.658–0.672)-0.044(-0.053, -0.034)-0.016(-0.017, -0.003)-0.179(-0.197, -0.037) Basic factors + TyG-WC0.693(0.686-0.700)-0.016(-0.025, -0.006)-0.004(-0.006, -0.001)-0.077(-0.096, -0.014) Basic factos + TG/HDL0.661(0.654–0.668)-0.048(-0.057, -0.038)-0.018(-0.020, -0.004)-0.196(-0.213, -0.048) Basic factors + CTI0.679(0.672–0.686)-0.030(-0.039, -0.020)-0.011(-0.013, -0.002)-0.123(-0.149, -0.028) Basic factors + SPISE0.680(0.673–0.687)-0.029(-0.038, -0.019)-0.011(-0.013, -0.002)-0.136(-0.152, -0.026) Basic factors + METS-IR0.694(0.687-0.700)-0.015(-0.025, -0.006)-0.003(-0.005, 0.002)-0.063(-0.080, -0.005) Basic factors + LAP0.669(0.662–0.676)-0.040(-0.049, -0.030)-0.015(-0.018, -0.003)-0.189(-0.201, -0.044)* Basic factors including age, sex, ethnicity. Abbreviations: TyG, triglyceride-glucose index; TyG-WC, TyG- waist circumference; TG/HDL, triglyceride-to-HDL cholesterol ratio; CTI, C-reactive protein-triglyceride-glucose index; SPISE, single-point insulin sensitivity estimator index; METS-IR, metabolic score for insulin resistance; LAP, lipid accumulation product Comparative Predictive Performance of Insulin Resistance Indices for Incident Acute Kidney Injury in Diabetes and Prediabetes * Basic factors including age, sex, ethnicity. Abbreviations: TyG, triglyceride-glucose index; TyG-WC, TyG- waist circumference; TG/HDL, triglyceride-to-HDL cholesterol ratio; CTI, C-reactive protein-triglyceride-glucose index; SPISE, single-point insulin sensitivity estimator index; METS-IR, metabolic score for insulin resistance; LAP, lipid accumulation product Sensitivity analyses To ensure the robustness of our primary findings, we conducted comprehensive sensitivity analyses with consistent results. First, the exclusion of participants with pre-existing CVD or CKD did not substantially alter the observed associations (Table S2), indicating these comorbidities did not significantly modify the relationships. Second, to address potential reverse causation, we excluded incident AKI cases occurring within the initial year, the first two years, the first three years, and the first five years of follow-up, which yielded materially unchanged associations (Table S3, Table S4, Table S5, Table S6). Third, further adjustment for key laboratory parameters (including HDL-C, LDL-C, fasting glucose, and serum albumin) and cardiovascular medications (antihypertensive and lipid-lowering therapies) similarly maintained the findings (Table S7). Finally, application of the updated hypertension definition (blood pressure ≥ 130/80 mmHg) produced consistent results (Table S8), further supporting the stability of our conclusions. Discussion This large-scale prospective cohort study provides stong evidence that lower eGDR, reflecting more severe IR, is independently associated with an elevated risk of AKI in individuals with diabetes or prediabetes. Notably, the association followed a dose-response pattern, with the highest eGDR quartile exhibiting a 62% lower AKI risk compared to the lowest quartile. Furthermore, eGDR demonstrated superior predictive performance for AKI risk compared to conventional IR indices, including the TyG index, TG/HDL ratio, CTI, SPISE, METS-IR and LAP. These findings highlight the potential utility of eGDR as a clinically accessible tool for identifying high-risk individuals who may benefit from targeted interventions to mitigate AKI risk. eGDR and AKI Risk Our study extends prior research by establishing a robust, inverse association between eGDR and AKI risk in a population with dysglycemia. The observed dose-response relationship suggests that even modest improvements in insulin sensitivity may confer renal protection. Mechanistically, IR contributes to AKI through multiple pathways, including endothelial dysfunction, impaired renal microcirculation, and heightened oxidative stress [34–40]—processes that are collectively captured by eGDR’s composite formula incorporating hypertension, abdominal adiposity, and HbA1c. Importantly, the association persisted after rigorous adjustment for established AKI risk factors, such as CKD, CVD, and inflammation, underscoring IR as an independent contributor to AKI pathogenesis. Translating these findings into clinical practice requires practical guidance on how to interpret eGDR values for risk stratification. Although eGDR demonstrates a linear dose-response relationship with AKI risk without a single definitive threshold, its clinical utility can be enhanced by risk stratification based on quartiles or clinically convenient cut-points. For instance, in our cohort, an eGDR value < 4.9 mg/kg/min (the lowest quartile) identifies individuals with an approximately 2.6-fold higher risk of AKI compared to those with eGDR ≥ 8.2 mg/kg/min (the highest quartile; adjusted HR 0.38, 95% CI: 0.34–0.43), and may serve as a practical and easily memorable ‘high-risk’ threshold. Conversely, eGDR ≥ 8.2 mg/kg/min denotes a low-risk phenotype. For intermediate values, the risk increases continuously as eGDR decreases, supporting a gradient-based risk communication approach. Clinicians may consider closer monitoring or preventive measures for individuals with eGDR values in the lower range, even if they do not fall below the absolute high-risk cut-point. Comparative predictive performance of eGDR A novel aspect of our study was the head-to-head comparison of eGDR with other IR indices. The superior discriminative accuracy of eGDR (C-index: 0.709) likely stems from its incorporation of clinical parameters (hypertension, waist circumference) alongside metabolic dysregulation (HbA1c), offering a more holistic reflection of IR’s systemic effects. In contrast, indices like TyG and TG/HDL rely solely on lipid-glucose metabolism, potentially overlooking the vascular and inflammatory components of IR that are particularly relevant to AKI. The significant net NRI and IDI with eGDR further supports its clinical utility for risk stratification. Stratified analyses and consistency Subgroup analyses revealed broadly consistent associations across demographic and clinical strata, reinforcing the robustness of our findings. While statistically significant interactions were observed for sex, BMI, and smoking status, the directionality of associations remained uniform, suggesting that the eGDR-AKI relationship is generalizable. Notably, the association was stronger in individuals without baseline glomerular or renal tubulo-interstitial diseases and CVD, possibly reflecting the compounding effects of glomerular or renal tubulo-interstitial diseases and CVD on AKI risk independent of IR. Clinical implications Our findings have important implications for AKI prevention in high-risk populations. First, eGDR—a readily calculable metric—could serve as a practical tool for identifying individuals with dysglycemia who are at heightened AKI risk, enabling closer monitoring or preemptive measures (e.g., nephrotoxic medication avoidance, hydration protocols). Second, the strong association between IR and AKI underscores the importance of lifestyle and pharmacological interventions aimed at improving insulin sensitivity, such as weight loss, physical activity, and metformin therapy, which may concomitantly reduce AKI risk. Limitations Several limitations warrant consideration. First, AKI cases were identified via ICD-10 codes rather than KDIGO criteria based on serial serum creatinine or urine output, as the UK Biobank does not capture real-time laboratory data at the time of AKI events. This approach may underestimate mild AKI cases and preclude severity staging. Future studies linking biobank data with electronic health records containing serial laboratory measurements are needed. Second, residual confounding from unmeasured factors (e.g., diet, genetic predisposition) cannot be entirely excluded. Third, the study population was predominantly White, and the relatively high mean BMI (29.5 kg/m²) may limit generalizability to lower-obesity populations. Nevertheless, a previous study in non-obese Chinese adults (mean BMI 23 kg/m²) demonstrated that several insulin resistance indices strongly predicted kidney disease progression [41], suggesting that insulin resistance contributes to renal risk independently of BMI. Further validation in diverse populations is warranted. Fourth, the UK Biobank lacks detailed baseline data on large vessel lesions (e.g., renal artery stenosis), which may influence AKI risk. We adjusted for prevalent CVD as a surrogate for macrovascular disease, but residual confounding cannot be entirely excluded. Fifth, while we excluded participants with pre-existing glomerular or tubulo-interstitial diseases and performed sensitivity analyses excluding early AKI cases up to five years, the observational design precludes definitive causal inference. Sixth, the eGDR formula was originally derived in type 1 diabetes populations. Although it has been widely validated, its coefficients may not be optimally weighted for prediabetic or older populations. Future studies incorporating clamp-derived insulin sensitivity measures are needed to re-evaluate and potentially recalibrate eGDR for specific populations. Furthermore, while we adjusted for the use of nonsteroidal anti-inflammatory drugs as a major nephrotoxic exposure in the community, the UK Biobank lacks comprehensive data on other potential nephrotoxins. Residual confounding from these unmeasured exposures cannot be entirely excluded. Finally, as an observational study, causality cannot be inferred, though the temporal sequence (baseline eGDR preceding AKI) strengthens our conclusions. Future large-scale prospective studies in diverse populations are needed to externally validate our findings. Furthermore, randomized controlled trials should explore whether IR-targeted therapies (e.g., lifestyle interventions, metformin, thiazolidinediones) attenuate AKI incidence in high-risk groups identified by low eGDR. Such studies would confirm the clinical validity and utility of eGDR as both a risk stratification tool and a potential therapeutic target. Conclusions In this large prospective cohort, lower eGDR was independently associated with higher AKI risk in individuals with diabetes or prediabetes, with a dose-response gradient. eGDR outperformed conventional IR indices in AKI risk prediction, highlighting its potential as a clinical tool for risk stratification. These findings advocate for further research into IR-modifying interventions as a strategy to reduce the growing burden of AKI in high-risk populations. Supplementary information Supplementary material 1. Supplementary material 1.