Bringing doctors together

or

A free, private network for verified physicians

Leading voices

Dr. Bruno Bezerra Lima

Advanced Cardiac Imaging · VUMC

View

Dr. Sara Ersozlu

Advanced Cardiac Imaging · MGH

View

Dr. Christopher Kramer

Advanced Cardiac Imaging · University of Virginia

View

Dr. Filipe Moura

Advanced Cardiac Imaging · Yale College

View

Dr. Robert Hendel

Advanced Cardiac Imaging · Tulane University

View

Dr. Venkatesh Murthy

Advanced Cardiac Imaging · UMHS

View

Dr. Dorairaj Prabhakaran

General Cardiology

View

Dr. Pugazhendhi Vijayaraman

EP · Geisinger

View

Dr. Robert Schaller

EP · HUP

View

Dr. Trey Woods

General Cardiology · Emory Healthcare

View
  • Discover your community

  • Stay current

  • A profile that represents you

  • Find your people

About RoonPrivacyTerms

    Overview

    Glucose Disposal Rate: A Novel Measure of Insulin Resistance Associated With Myocardial Fibrosis and Incident Heart Failure.

    May 6, 2026DOI: 10.1161/jaha.125.047908Journal ArticleAHA

    Written by: Syed BukhariYaa A KwapongLisa R YanekErin D MichosJoao A C LimaWendy S PostCynthia C TaubKavita SharmaMonica MukherjeeBenjamin H FreedAnum S MinhasElnaz EbrahimihoorJustin B Echouffo-TcheuguiAllison G Hays

    Find these authors on Roon

    BF
    Dr. Benjamin Freed
    Advanced Cardiac Imaging · Northwestern Medicine

    METHODS

    We used data from MESA (Multi‐Ethnic Study of Atherosclerosis), which are available through the National Heart, Lung, and Blood Institute's Biologic Specimen and Data Repository. The study design and methods of the MESA study have been previously described.14 MESA is a diverse community‐based sample of 6814 men and women aged 45 to 84 years, free of apparent clinical cardiovascular disease and HF at baseline in 2000 to 2002. Participants were drawn from 6 diverse locations across the United States: New York, NY; Baltimore, MD; Chicago, IL; Los Angeles, CA; St. Paul, MN; and Winston‐Salem, NC. The first visit for participants (visit 1) occurred between 2000 and 2002 and was followed by subsequent visits. During each visit, data collection included demographic information, medical history, physical assessments, and medication use, as previously documented.14We included participants from visit 2, because HbA1c needed to calculate the exposure of interest (eGDR) was measured at visit 2 and not visit 1. We excluded participants with missing data on the exposure (n=153), prevalent HF (n=47), and those with no follow‐up (n=8). A total of 6025 participants were included in our final cohort (Figure 1).After excluding participants without exposure or outcome, several more participants (n=8) had no follow‐up and no event data and so had to be further excluded. eGDR indicates estimated glucose disposal rate; and HF, heart failure.The MESA protocol was approved by the institutional review board at each MESA participating site. All participants provided written informed consent. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for observational studies.The independent variable is insulin resistance defined by eGDR using the formula: eGDR = 19.02 − (0.22 × BMI, in kilograms divided by height in meters squared) − (3.26 × hypertension, presence) − (0.61 × HbA1c %). BMI was calculated from weight and height (kilograms divided by height in meters squared), measured using a standardized protocol.8 HbA1c was measured using high‐performance liquid chromatography (Tosoh G7; Tosoh Corporation, Tokyo, Japan). Hypertension was defined as systolic BP ≥130 mm Hg or diastolic BP ≥80 mm Hg, or self‐reported antihypertensive medication use. The resting blood pressure of participants was obtained after a 5‐minute rest in the seated position using a Dinamap Pro 100 automated oscillometric sphygmomanometer operated by a trained technician.
    The primary outcome of interest was incident HF events that occurred after baseline (MESA exam 2) through the year 2020, the time of last event adjudication completion. The process of event ascertainment in MESA has been previously described.14 Briefly, participants were contacted by telephone every 9 to 12 months to obtain information on interim hospitalizations. If hospitalization was reported, medical records were obtained. A panel of trained MESA physicians formally adjudicated the diagnosis of HF hospitalization using standardized criteria based on a review of medical records. The study included both probable and definite HF events.15, 16 Probable HF was defined as a physician diagnosis or HF treatment. A definite HF diagnosis required at least 1 additional objective measure such as radiographic evidence of reduced LV function by echocardiography or ventriculography, or findings consistent with LV diastolic dysfunction. HFrEF was defined as an HF event with an ejection fraction of <45%, whereas HFpEF was defined as an HF event with ejection fraction ≥45%.
    MESA participants completed a CMR imaging study at MESA exams 1 and 5. We used CMR imaging data from eligible participants at exam 5. The MESA CMR imaging study inclusion and exclusion criteria and imaging protocols have been previously described.17 CMR imaging was conducted using 1.5T scanners (Avanto and Espree, Siemens Healthineers; Signa HD, GE Healthcare) equipped with 6‐channel anterior phased‐array coils and electrocardiographic gating. Late gadolinium enhancement images were obtained ≈15 minutes following intravenous administration of a gadolinium‐based contrast agent (0.15 mmol/kg of Magnevist; Bayer Healthcare Pharmaceuticals). Imaging included 10 to 12 contiguous short‐axis slices encompassing the entire left ventricle, along with horizontal and vertical long‐axis views. The 2‐dimensional biplane late gadolinium enhancement technique used is a clinical standard for assessing LV scar, and imaging procedures were standardized across all participating centers. Myocardial fibrosis was defined as hyperenhancement in either 1 short‐axis and 1 long‐axis image at a matching location or 2 adjacent short‐axis slices.17
    Secondary outcomes included evidence of myocardial dysfunction on CMR imaging, specifically LV ejection fraction and the presence of myocardial fibrosis. A total of 1780 participants were included for the analysis of myocardial fibrosis by CMR imaging, and a total of 2899 for LV ejection fraction, because not all participants underwent late gadolinium enhancement on CMR imaging.
    Covariates were assessed at visit 2 in MESA by standard questionnaires, physical examination, and laboratory tests. These included sociodemographic characteristics such as self‐reported age (in years), race and ethnicity, sex, and education. Smoking, physical activity, medical history and medication use were self‐reported through standardized questionnaires. Plasma total cholesterol and high‐density lipoprotein cholesterol (HDL‐C) concentrations were measured using standard enzymatic methods. Physical activity was assessed using a 28‐item Typical Week Physical Activity Survey and estimated in metabolic equivalent minutes per week.18
    Participants' baseline characteristics overall and by eGDR tertiles were summarized using descriptive statistics, including proportions for categorical variables and means±SDs or medians (interquartile ranges) for continuous variables. These characteristics were compared among eGDR tertiles using χ2 tests for categorical variables. Continuous variables were assessed for normality using the Shapiro‐Wilk test. Variables that had approximately normal distributions and passed the Brown‐Forsythe homogeneity of variance test were compared across eGDR tertiles using 1‐way ANOVA, whereas variables with skewed distributions that did not pass the Brown‐Forsythe homogeneity of variance test were compared across eGDR tertiles using the Kruskal‐Wallis test.
    The rates of HF (overall and by HF subtypes) were calculated as the number of adjudicated cases observed per 1000 person‐years at risk. The person‐time follow‐up was from the baseline examination (visit 2 until the first occurrence of HF) or death. HF free survivor curves were plotted with Kaplan‐Meier curves. Multivariable Cox proportional hazards regression models were used to examine the association of eGDR at exam 2 (as a continuous variable) with incident HF in a series of progressively adjusted models: Model 1 was unadjusted; Model 2 was adjusted for age, sex, and race and ethnicity; Model 3 included Model 2 variables plus MESA field site, education; Model 4 included Model 3 variables plus smoking, physical inactivity, total cholesterol, HDL‐C, use of lipid lowering medications; Model 5 included Model 4 variables plus urine albumin/creatinine. Covariates were selected based on a conceptual framework informed by prior literature on insulin resistance and HF, established modeling approaches in previous MESA analyses, and causal inference principles using a directed acyclic graph approach.15, 19, 20, 21, 22, 23 Variables were considered confounders if they were hypothesized to be common causes of both eGDR and incident HF and were not on the presumed causal pathway. Age, sex, and race and ethnicity were included a priori as fundamental demographic determinants of both insulin resistance and HF risk. Smoking status, physical inactivity, total cholesterol, HDL‐C, and lipid‐lowering medication use were included as established cardiovascular risk factors associated with both metabolic dysfunction and HF risk. The MESA field site was included to account for geographic and recruitment‐related variation comparable with prior MESA analyses that adjusted for study site.15, 23 Because eGDR is calculated using BMI and hypertension status, these variables were not included as covariates to avoid overadjustment bias and mathematical coupling with the exposure. We used a stepwise, progressive modeling strategy to evaluate the robustness of associations with sequential adjustment for demographic, socioeconomic, site‐related, and cardiometabolic risk factors. This approach allowed us to assess the extent to which observed associations were explained by potential confounders. This hierarchical modeling strategy is consistent with established epidemiologic practice for assessing incremental confounding and stability of associations.19
    The regression diagnostic metrics were examined; all models showed no collinearity and met assumptions for proportional hazards; thus, nonrobust standard errors were used. Interaction between eGDR and each covariate was tested by addition of an interaction term to Model 5.
    Multivariable linear regression models were used to examine the association of eGDR at exam 2 (continuous variable) with LV ejection fraction from CMR imaging at exam 5 in a series of progressive models: (1) unadjusted; (2) adjusted for age, sex, and race and ethnicity; (3) Model 2 plus MESA field site, education; (4) Model 3 plus smoking, physical inactivity, total cholesterol, HDL‐C, use of lipid lowering medications; (5) Model 4 plus urine albumin/creatinine. Regression diagnostic metrics were examined for all models; residuals were normally distributed and no collinearity, variance inflation, or undue influence was observed; thus, nonrobust standard errors were used.
    Multivariable logistic regression models were used to investigate the association of eGDR at exam 2 (continuous) with LV fibrosis (yes/no) at exam 5 in a series of progressive models: (1) unadjusted; (2) adjusted for age, sex, and race and ethnicity; (3) Model 2 plus MESA field site, education; (4) Model 3 plus smoking, physical inactivity, total cholesterol, HDL‐C, use of lipid lowering medications. All models passed goodness‐of‐fit tests with no collinearity observed; thus, nonrobust standard errors were used.
    All models were stratified by diabetes status, categorized as normoglycemia, prediabetes (defined as having a fasting glucose level between 100 and 125 mg/dL), and diabetes (fasting glucose level of 126 mg/dL or higher, or the use of glucose‐lowering medications), to assess whether associations between eGDR and outcomes differed across glycemic states. A 2‐sided α level of <0.05 was used to assess the statistical significance of the results. All analyses were performed from November 2024 to February 2025 using SAS version 9.4 (SAS Institute, Cary, NC) and Stata version 16 (StataCorp, College Station, TX).

    RESULTS

    Baseline characteristics of participants overall and by tertiles of eGDR are described in Table 1. Among the 6025 participants (mean age, 63.6±10.1 years), 52.4% were female adults, and 47.6% were male adults. Overall, ≈40% self‐identified as non‐Hispanic White adults, 26.7% as non‐Hispanic Black adults, 22% as Hispanic adults, and 11.9% as adults of Chinese ancestry. The majority (63%) of participants reported having less than a bachelor's degree level of education. The median BMI of participants was 27.6 kg/m2, and median HbA1c was 5.5%.
    Baseline (Exam 2) Characteristics of Participants Overall and by Tertiles of Estimated Glucose Disposal Rate (n=6025)
    ACE indicates angiotensin‐converting enzyme; ADA, American Diabetes Association; ARB, angiotensin receptor blocker; HbA1c, glycated hemoglobin; HDL, high‐density lipoprotein; LDL, low‐density lipoprotein; and MET, metabolic equivalent.
    P values could not be calculated because the variables are involved in the calculation of estimated glucose disposal rate.
    The eGDR among participants ranged from −1.99 to 12.33, with a median of 7.93 (interquartile range, 6.02–9.90). Participants within the lowest tertile (T1) of eGDR (indicating highest insulin resistance) were more likely to have treated diabetes compared with T2 and T3 (24.3% versus 9.0% versus 1.9%, P<0.001, respectively). Compared with participants in T2 and T3, participants in T1 were older (T1: 65.8 years, T2: 63.8 years, T3: 61.1 years; P<0.001), more likely to use antihypertensive medications (T1: 80.6%, T2: 38.7%, T3: 5.9%; P<0.001), and had higher aspirin use (T1: 39%, T2: 31.4%, T3: 21.8%; P<0.001).
    Over a mean follow‐up period of 14±5 years, there were 404 HF events: 200 were HFpEF and 168 were HFrEF, whereas 36 events were unclassified/indeterminate. Overall, the event rate for HF was 4.67 per 1000 person‐years, whereas that of HFpEF and HFrEF were 2.32 per 1000 person years and 1.95 per 1000 person years, respectively.
    HF‐free survival curves by tertiles of eGDR for overall HF, HFpEF, and HFrEF are displayed in Figure 2. For overall HF, participants in the lowest eGDR tertile (T1, indicating highest insulin resistance) had a significantly higher cumulative incidence of HF compared with those in tertiles 2 and 3 (log‐rank P<0.001). A similar pattern was observed in both HFpEF and HFrEF, where the risk was highest in participants in the lowest eGDR tertile.
    eGDR indicates estimated glucose disposal rate.
    Lower eGDR (indicative of higher insulin resistance) was significantly associated with higher risk of overall incident HF in the unadjusted model (hazard ratio [HR] per each 1‐unit lower eGDR: 1.26 [95% CI, 1.22–1.31]) and when adjusted for age, sex, and race and ethnicity (adjusted HR per each 1‐unit lower eGDR, 1.29 [95% CI, 1.23–1.35]) (Table 2). When additionally adjusted for education, MESA field site, smoking, physical inactivity, total cholesterol, HDL‐C, and use of lipid lowering medications, the significant association persisted (adjusted HR per each 1‐unit lower eGDR, 1.28 [95% CI, 1.22–1.35]). In sensitivity analyses stratified by diabetes status (normoglycemia, prediabetes, and diabetes), the association remained significant, indicating that eGDR predicts risk of HF even among individuals without diabetes (Table 2). The association was stronger for HFpEF (age‐, sex‐, race‐adjusted HR per each 1‐unit lower eGDR: 1.36 [95% CI, 1.27–1.45]) than HFrEF (HR, 1.19 [95% CI, 1.11–1.27]). No statistically significant interaction effects were observed (Table S1).
    Association of Estimated Glucose Disposal Rate at Exam 2 (Continuous) With Incident Heart Failure Through 2020, Overall and Stratified by Diabetes Status
    HDL‐C indicates high‐density lipoprotein‐cholesterol; HR, hazard ratio; MESA, Multi‐Ethnic Study of Atherosclerosis; and UACr, urine albumin/creatinine.
    Lower levels of eGDR were inversely associated with LV ejection fraction in unadjusted and adjusted models. In the fully adjusted Model 4, per 1‐unit decrease in eGDR the LV ejection fraction was 0.16% higher (n=2899; β=0.16 per 1‐unit change in eGDR [95% CI, 0.03–0.28]) (Table 3). Furthermore, lower eGDR was significantly associated with higher odds of LV fibrosis in all the models. Each 1‐unit decrease in eGDR was associated with a 27% higher odds of later‐life prevalent LV fibrosis, holding all adjusted variables constant (fully adjusted Model 4: n=1780; odds ratio, 1.27 [95% CI, 1.16–1.41]).
    Association of Estimated Glucose Disposal Rate at Exam 2 With Cardiac Function (on Cardiac Magnetic Resonance Imaging) at Exam 5
    β indicates absolute difference; eGDR, estimated glucose disposal rate; HDL‐C, high‐density lipoprotein cholesterol; LV, left ventricular; and MESA, Multi‐Ethnic Study of Atherosclerosis.

    DISCUSSION

    In this large, multiethnic, community‐based cohort of adults, we demonstrated that lower eGDR, reflecting higher insulin resistance, is independently associated with an increased risk of incident HF. This association persisted after adjustment for multiple demographic, clinical, and metabolic risk factors, remained consistent across diabetes strata, and was observed in both HFpEF and HFrEF. These findings highlight the potential of eGDR as a simple, accessible marker of HF risk even in patients without diabetes or prediabetes. Beyond clinical HF, lower eGDR was also associated with increased myocardial fibrosis, suggesting that insulin resistance may contribute to early cardiac remodeling.
    Our results extend prior evidence linking metabolic disturbances to adverse cardiovascular outcomes. Although traditional risk factors such as obesity, hypertension, and diabetes are well‐established contributors to HF development, growing evidence suggests that insulin resistance plays a central role in myocardial remodeling and dysfunction, even in the absence of ≥1 of these risk factors.24, 25 Prior population‐based studies, such as the ULSAM (Uppsala Longitudinal Study of Adult Men) and ARIC (Atherosclerotic Risk in Communities) cohorts, have demonstrated that higher homeostasis model assessment of insulin resistance levels are associated with incident HF5, 21; however, these studies did not assess subclinical myocardial damage and also did not assess HF subtypes. Our study now demonstrates that eGDR, a validated estimate of insulin resistance derived from routinely available clinical variables, is associated with future HF events (including both HFpEF and HFrEF) and myocardial fibrosis, providing a more feasible tool for risk stratification in clinical practice.
    Several mechanisms may explain the association between insulin resistance, clinical HF, and myocardial fibrosis. The myocardium normally responds to injury by altering substrate metabolism to increase energy efficiency. In insulin resistance, this adaptive response is inhibited, because the myocardium uses free fatty acids and less glucose, and the metabolic irregularity ultimately contributes to lipotoxicity, sympathetic upregulation, inflammation, oxidative stress, and fibrosis.26 Experimental data in animal models of diabetes have shown consistent evidence of cardiac hypertrophy, fibrosis, and diastolic dysfunction, whereas human studies demonstrate that patients with diabetes, even when asymptomatic, often exhibit diffuse myocardial fibrosis and impaired relaxation.27 Our current findings are consistent with this paradigm and suggest that eGDR may capture early metabolic derangements that precede overt HF regardless of diabetes status.
    It is important to mention that although diabetes is a well‐established risk factor for HF, it reflects only 1 aspect of cardiometabolic dysfunction. In contrast, eGDR provides a dynamic and quantitative assessment of insulin resistance by integrating HbA1c, BMI, and hypertension, 3 interrelated risk factors that each contribute independently to HF risk. As such, eGDR captures a broader spectrum of dynamic metabolic processes than diabetes alone, offering a more comprehensive risk assessment tool. This added granularity allows eGDR to stratify risk even among individuals without diagnosed diabetes (but possibly with dysglycemia) and may enhance its value as both a predictive and surveillance marker for HF.
    Notably, in our study, the association between lower eGDR and HF was observed for both HFpEF and HFrEF, with a somewhat stronger signal for HFpEF. This aligns with prior work showing that metabolic syndrome, microvascular dysfunction, and systemic inflammation are prominent contributors to HFpEF pathophysiology.28, 29 This suggests that early metabolic derangements, particularly insulin resistance, may impair myocardial relaxation and contribute to myocardial stiffness, eventually manifesting as HFpEF. Meanwhile, associations with HFrEF may reflect the long‐term effects of insulin resistance on cardiac energetics, mitochondrial function, and structural remodeling.28 Our findings suggest that insulin resistance may represent a shared upstream driver of both HF subtypes, potentially through pathways involving myocardial fibrosis and impaired energetics.28
    Beyond clinical HF, our findings of an inverse relationship between baseline eGDR and myocardial fibrosis at follow‐up provide further evidence of subclinical myocardial damage in those with greater insulin resistance. Fibrosis is an important pathophysiologic hallmark of HF and is thought to result from a cascade of inflammation, oxidative stress, and altered myocardial substrate usage, all of which have been linked to insulin resistance.26 These findings are clinically meaningful, because subclinical myocardial fibrosis has been linked to adverse outcomes30, 31 and may represent an early pathophysiological stage preceding the onset of overt clinical HF.32 Similarly, the inverse relationship between eGDR and LV ejection fraction likely reflects the impact of insulin resistance on adverse cardiac remodeling that preferentially predisposes individuals to HFpEF than HFrEF.
    Importantly, our findings suggest that incorporating eGDR into routine risk stratification could help identify individuals at higher risk for HF earlier in the disease trajectory, potentially allowing for targeted preventive strategies focused on improving insulin sensitivity. Lifestyle interventions and pharmacologic therapies such as GLP‐1 (glucagon‐like peptide‐1) receptor agonists and SGLT2 (sodium‐glucose cotransporter 2) inhibitors, which improve metabolic parameters and reduce cardiovascular risk,33, 34 may offer particular benefit in this context. Additionally, pioglitazone, a thiazolidinedione that enhances insulin sensitivity, is a consideration, but its potential use is limited by the risk of fluid retention. Furthermore, it would be worthwhile investigating whether serial monitoring of eGDR could provide prognostic information and guide preventative strategies in these patients.
    Our study has several limitations. First, eGDR was assessed at a single time point, and we could not evaluate the impact of changes in insulin resistance over time. Second, although eGDR provides a practical estimate of insulin sensitivity, it remains an indirect estimate and may not fully capture the dynamic insulin‐glucose interactions. Furthermore, because eGDR incorporates hypertension, which is itself a potent and independent risk factor for HF via non‐insulin resistance mechanisms, the observed association between lower eGDR and HF may partly reflect the contribution of hypertension rather than insulin resistance alone. Nevertheless, eGDR correlates closely with gold‐standard measures of insulin resistance. Last, as an observational analysis, causality cannot be inferred in the present analysis, and residual confounding may persist despite multivariable adjustment.

    CONCLUSIONS

    Lower eGDR, indicating greater insulin resistance, is independently linked to a higher risk of developing HF and to greater myocardial fibrosis. These results highlight insulin resistance as a potential driver in the pathogenesis of both HFpEF and HFrEF, and position eGDR as a promising tool for early risk stratification and targeted prevention. Future studies should explore whether interventions that improve insulin sensitivity can mitigate HF risk, particularly in individuals with lower eGDR.

    Sources of Funding

    This work is supported by National Institutes of Health grants National Institutes of Health/National Heart, Lung, and Blood Institute T32HL007227 (SB) and National Institutes of Health/National Heart, Lung, and Blood Institute R35HL172680 (A.G.H.). The parent MESA study was supported by contracts 75N92020D00001, HHSN268201500003I, N01‐HC‐95159, 75N92020D00005, N01‐HC‐95160, 75N92020D00002, N01‐HC‐95161, 75N92020D00003, N01HC‐95162, 75N92020D00006, N01‐HC‐95163, 75N92020D00004, N01‐HC‐95164, 75N92020D00007, N01‐HC95165, N01‐HC‐95166, N01‐HC‐95167, N01‐HC‐95168, and N01‐HC‐95169 from the National Heart, Lung, and Blood Institute, and by grants UL1‐TR‐000040, UL1‐TR‐001079, and UL1‐TR‐001420 from the National Center for Advancing Translational Sciences.

    Disclosures

    None.

    Supporting information

    Table S1
    STROBE Checklist
    BFDr. Benjamin Freed

    Dr. Benjamin Freed

    Cardiology · Advanced Cardiac Imaging
    Northwestern Medicine
    benjaminfreed