Abstract
Background
Variability in kidney function, as measured by estimated glomerular filtration rate (eGFR), has been proposed as a novel risk factor for adverse cardiovascular outcomes. However, its association with incident heart failure (HF) in adults with type 2 diabetes remains unclear.
Objectives
The aim of the study was to evaluate the association of eGFR variability with incident HF in a large sample of adults with type 2 diabetes.
Methods
We included participants who had ≥5 eGFR measurements in the first 24 months of the ACCORD (Action to Control Cardiovascular Risk in Diabetes) study. The visit-to-visit variability of eGFR was estimated using the coefficient of variation (CV), SD, and variability independent of the mean. Cox proportional hazards models were used to estimate adjusted HRs and 95% CIs for incident HF.
Results
Among 7,601 participants (mean age: 63 ± 6.5 years, 39.1% women, 63% White), 193 developed HF over a median follow-up of 3 years. HF incidence increased across quartiles of eGFR variability. In fully adjusted models, participants in the highest quartile of eGFR CV had an HR of 1.88 (95% CI: 1.09-3.22; P trend = 0.019) compared to the lowest quartile. Per 1-SD increments in all the eGFR variability metrics were associated with increased HF risk (CV: HR: 1.15; 95% CI: 0.99-1.33; SD: HR: 1.16; 95% CI: 1.01-1.33; variability independent of the mean: HR: 1.18; 95% CI: 1.03-1.35).
Conclusions
Higher variability in eGFR is independently associated with an increased risk of incident HF in adults with type 2 diabetes.
Key words: heart failure, kidney function, type 2 diabetes, variability
Central Illustration
Heart failure (HF) poses a significant public health challenge, particularly among individuals with type 2 diabetes mellitus (T2DM), where its prevalence and associated burden are disproportionately high.1, 2, 3 Decreased kidney function, as reflected by reduced estimated glomerular filtration rate (eGFR), is a well-established risk factor for HF in both the general population and specifically among individuals with T2DM.4, 5, 6, 7, 8 Impaired renal function contributes to HF risk through mechanisms such as fluid overload, activation of the renin-angiotensin-aldosterone system (RAAS), and heightened systemic inflammation, which are particularly pronounced in people with T2DM.9
Variability in kidney function, as assessed by fluctuations in eGFR, has emerged as a potential marker of cardiovascular risk.10 While a single eGFR measurement provides valuable insights into renal health, longitudinal variations in eGFR may capture dynamic pathophysiologic processes such as hemodynamic instability, subclinical inflammation, and adverse metabolic changes, all of which may contribute to increased cardiovascular risk.9,10 Previous studies have linked eGFR variability to adverse atherosclerotic cardiovascular outcomes.9 However, the relationship between eGFR variability and the risk of incident HF, particularly among individuals with T2DM, has not been investigated. Clarifying this relationship is crucial given the shared pathophysiologic pathways between kidney dysfunction and HF among individuals with T2DM.9 Both conditions are characterized by overlapping mechanisms, including RAAS activation, sympathetic nervous system overactivity, endothelial dysfunction, and chronic inflammation, which promote adverse cardiac remodeling and progressive organ damage.9 Understanding the link between eGFR variability and HF may provide novel insights into cardiovascular risk stratification in T2DM.
In this study, we investigated the association between eGFR variability and incident HF among participants with T2DM in the ACCORD (Action to Control Cardiovascular Risk in Diabetes) trial. We hypothesized that higher eGFR variability would be independently associated with an increased risk of incident HF, even after accounting for established cardiovascular risk factors and baseline kidney function.
Methods
Study design
This study is a secondary analysis of data from the ACCORD trial, a multicenter, randomized controlled trial conducted from 2001 to 2005 at 77 sites across the United States and Canada.11 The trial enrolled 10,251 participants aged 40 to 79 years with either established cardiovascular disease (CVD) or, for those aged 55 to 79 years, additional risk factors such as albuminuria, atherosclerosis, left ventricular hypertrophy, or at least 2 other CVD risk factors. Participants were randomly assigned to interventions targeting intensive or standard glucose control. Subgroups were further randomized to lipid-lowering or blood pressure interventions. A detailed description of the ACCORD trial design and objectives is available in prior publications.11
For this analysis, we excluded individuals with less than 5 eGFR data points (n = 1,140), HF at baseline (n = 317), and those who did not complete the 24-month eGFR variability assessment period (n = 173). We also excluded participants with a history of fibrate use (n = 1,020), as fibrates have been shown to increase serum creatinine without causing any alterations in kidney function.12 A summary of the exclusion process is provided in the study flow diagram (Supplemental Figure 1).
All participants provided written informed consent, and the study protocol received approval from the institutional review boards at participating institutions.
Assessment of eGFR variability
At each study visit and location, blood samples were collected from all participants and analyzed on the same day they were received. Serum creatinine levels were determined using enzymatic methods on a Roche Double Modular P Analytics automated analyzer. The interassay precision coefficients were reported to be below 1.4% for high-quality control samples and below 2.2% for low-quality control samples.13 eGFR was calculated in mL/min/1.73 m2 using the Chronic Kidney Disease Epidemiology Collaboration equation.14 To quantify visit-to-visit variability in eGFR, 3 metrics were employed: 1) coefficient of variation (CV); 2) intraindividual SD; and 3) variability independent of the mean (VIM). VIM was calculated using the formula 100∗SD/meanβ, where β represents the regression coefficient derived from the natural logarithm of SD as a function of the natural logarithm of the mean eGFR. The use of several metrics was done to comprehensively capture fluctuations in renal function.
Ascertainment of incident heart failure events
HF events were identified through participant reports during scheduled clinic visits every 4 months, supplemented by telephone follow-up for missed visits. Events were defined as HF-related hospitalizations or deaths attributed to HF, confirmed through clinical imaging and adjudication processes by an expert committee. Participants were followed until the first occurrence of HF, death, or the trial’s conclusion in 2009.
Covariates
The covariates assessed at baseline included age, sex, race/ethnicity, treatment assignments, smoking status, alcohol use, systolic blood pressure (BP), diastolic BP, duration of diabetes, history of atherosclerotic cardiovascular disease (ASCVD), and medication use, including use of insulin/sulfonylurea, thiazolidinediones, angiotensin-converting enzyme inhibitor (ACEi)/angiotensin receptor blockers (ARBs). Left ventricular hypertrophy was defined by Cornell voltage criteria on the baseline electrocardiogram. Additional covariates included time-averaged measures for body mass index, systolic BP and hemoglobin A1C, total/high-density lipoprotein cholesterol ratio, and urinary albumin-creatinine ratio calculated from all available visits during the eGFR variability assessment period. Urine albumin was measured from spot urine samples via immunonephelometry using a Siemens BN11 nephelometer, with a sensitivity of 0.16 mg/dL. The interassay coefficients of variation were 3.0%, 2.6%, and 4.9% for control levels of 0.89 mg/dL, 6.6 mg/dL, and 16.1 mg/dL, respectively. Urine creatinine levels were determined using enzymatic methods on a Roche Double Modular P Analytics automated analyzer.13 Urinary albumin excretion was expressed as the urine albumin-to-creatinine ratio (UACR) in mg of albumin per g of creatinine.
Statistical analyses
The study participants were categorized into quartiles based on eGFR variability metrics. Baseline characteristics were compared across quartiles using analysis of variance or Kruskal-Wallis tests for continuous variables and chi-square tests for categorical variables.
Incidence rates of HF were expressed per 1,000 person-years, calculated by dividing the number of HF events by total at-risk person-years. Cox proportional hazards models were used to estimate HRs and 95% CIs for HF risk, adjusting for potential confounders in sequential models. Model 1 adjusted for age, sex, race, and treatment arm; model 2 included model 1 plus duration of diabetes, cigarette smoking, alcohol intake, mean body mass index, mean systolic blood pressure, mean total/high-density lipoprotein cholesterol ratio, mean glycated hemoglobin, mean urinary albumin-creatinine ratio, left ventricular hypertrophy, use of insulin/sulfonylurea, use of thiazolidinediones, use of antihypertensive medication, use of ACEi/ARB, use of beta blocker, and history of prevalent ASCVD; model 3 included model 2 plus cardiac autonomic neuropathy; model 4 included model 3 plus incident coronary artery disease as a time-varying covariate; and model 5 included model 4 plus mean eGFR. In sensitivity analyses, we performed additional adjustments with model 6, including model 5 plus use of diuretics (loop, thiazide, and potassium-sparing diuretics), and model 7, including model 6 plus prevalent retinopathy at baseline.
We also conducted interaction analyses to examine whether the relationship between eGFR variability measures and incident HF differs across the treatment arms (including BP, lipid, and glycemic control treatment arms).
All analyses were performed using STATA version 14.2 (StataCorp LLC), with statistical significance set at P < 0.05.
Results
Characteristics of study participants
Table 1 shows the characteristics of study participants according to quartiles of eGFR variability. Compared to those in lower quartiles, participants in the highest quartile (Q4) of eGFR variability were older and more frequently Black or women. They also had longer duration of diabetes, lower eGFR, higher averages of body mass index, HbA1C, total/high-density lipoprotein cholesterol ratio, and UACR, as well as higher rates of ASCVD and use of insulin, beta blocker, and ACEi/ARB at baseline.
Table 1.
Characteristics of Participants by eGFR Variability
| Total (N = 7,601) | Coefficient of Variation of eGFR, % |
P Value | ||||
|---|---|---|---|---|---|---|
| Q1 (<6.23) (n = 1,901) |
Q2 (6.23-9.77) (n = 1,900) |
Q3 (9.78-14.57) (n = 1,900) |
Q4 (>14.57) (n = 1,900) |
|||
| At baseline | ||||||
| Age, mean (SD), y | 62.7 (6.5) | 61.3 (6.2) | 62.6 (6.6) | 63.2 (6.4) | 63.7 (6.6) | <0.001 |
| Sex | <0.001 | |||||
| Women | 2,974 (39.1) | 669 (35.2) | 732 (38.5) | 771 (40.6) | 802 (42.2) | |
| Men | 4,627 (60.9) | 1,232 (64.8) | 1,168 (61.5) | 1,129 (59.4) | 1,098 (57.8) | |
| Race and ethnicity | <0.001 | |||||
| White | 4,788 (63.0) | 1,249 (65.7) | 1,182 (62.2) | 1,159 (61.0) | 1,198 (63.1) | |
| Black | 1,448 (19.1) | 225 (11.8) | 379 (19.9) | 454 (23.9) | 390 (20.5) | |
| Hispanic | 524 (6.9) | 161 (8.5) | 120 (6.3) | 117 (6.2) | 126 (6.6) | |
| Other | 841 (11.1) | 266 (14.0) | 219 (11.5) | 170 (8.9) | 186 (9.8) | |
| Treatment arm | 0.012 | |||||
| Standard glycemic lowering | 3,832 (50.4) | 986 (51.9) | 966 (50.8) | 983 (51.7) | 897 (47.2) | |
| Intensive glycemic lowering | 3,769 (49.6) | 915 (48.1) | 934 (49.2) | 917 (48.3) | 1,003 (52.8) | |
| Current smoking | 1,013 (13.3) | 293 (15.4) | 259 (13.6) | 239 (12.6) | 222 (11.7) | 0.005 |
| Alcohol drinking | 1,885 (24.8) | 518 (27.2) | 480 (25.3) | 461 (24.3) | 426 (22.4) | 0.006 |
| Diabetes duration, median (IQR), y | 9.0 (5.0, 15.0) | 8.0 (4.0, 13.0) | 9.0 (5.0, 14.0) | 10.0 (5.0, 16.0) | 10.0 (5.0, 16.0) | <0.001 |
| Use of BP-lowering drug | 6,300 (82.9) | 1,456 (76.6) | 1,569 (82.6) | 1,612 (84.8) | 1,663 (87.5) | <0.001 |
| Use of beta-blocker | 2,090 (27.6) | 437 (23.0) | 515 (27.2) | 515 (27.2) | 623 (32.9) | <0.001 |
| Use of ACEI/ARB | 5,256 (69.1) | 1,221 (64.2) | 1,277 (67.2) | 1,355 (71.3) | 1,403 (73.8) | <0.001 |
| Use of insulin | 2,560 (33.7) | 518 (27.2) | 633 (33.3) | 681 (35.8) | 728 (38.3) | <0.001 |
| Use of sulfonylurea | 4,085 (53.8) | 1,019 (53.6) | 1,040 (54.7) | 1,010 (53.2) | 1,016 (53.5) | 0.79 |
| Use of thiazolidinediones | 1,701 (22.4) | 422 (22.2) | 398 (20.9) | 449 (23.6) | 432 (22.7) | 0.24 |
| LVH | 512 (6.7) | 100 (5.3) | 134 (7.1) | 132 (6.9) | 146 (7.7) | 0.021 |
| History of ASCVD | 2,385 (31.4) | 532 (28.0) | 587 (30.9) | 583 (30.7) | 683 (35.9) | <0.001 |
| Over eGFR assessment period | ||||||
| Number of eGFR measurements | <0.001 | |||||
| 5 | 3,864 (50.8) | 1,050 (55.2) | 965 (50.8) | 928 (48.8) | 921 (48.5) | |
| 6 | 1,018 (13.4) | 225 (11.8) | 270 (14.2) | 238 (12.5) | 285 (15.0) | |
| 7 | 2,719 (35.8) | 626 (32.9) | 665 (35.0) | 734 (38.6) | 694 (36.5) | |
| Average BMI, mean (SD), kg/m2 | 32.6 (5.5) | 32.4 (5.6) | 32.3 (5.5) | 32.7 (5.4) | 32.8 (5.6) | 0.008 |
| Average hemoglobin A1C, mean (SD), % | 7.3 (0.8) | 7.3 (0.8) | 7.3 (0.8) | 7.3 (0.8) | 7.4 (0.8) | 0.036 |
| Average SBP, mean (SD), mm Hg | 129.7 (12.2) | 129.4 (11.3) | 129.5 (11.6) | 130.1 (12.7) | 129.7 (13.1) | 0.30 |
| Average DBP, mean (SD), mm Hg | 71.0 (8.2) | 72.1 (8.0) | 71.3 (7.8) | 70.8 (8.5) | 69.7 (8.3) | <0.001 |
| Average TC/HDL ratio, mean (SD) | 4.3 (1.7) | 4.2 (1.5) | 4.2 (1.5) | 4.2 (1.5) | 4.5 (2.2) | <0.001 |
| Average UACR, median (IQR), mg/g | 13.0 (7.0, 40.0) | 11.5 (6.5, 30.0) | 11.5 (6.5, 29.0) | 14.0 (7.0, 43.5) | 17.0 (7.5, 68.0) | <0.001 |
| Average eGFR, mean (SD), mL/min/1.73 m2 | 79.3 (16.9) | 93.2 (12.6) | 82.6 (14.3) | 75.2 (14.1) | 66.3 (13.9) | <0.001 |
ACEI = angiotensin-converting enzyme inhibitor; ARB = angiotensin II receptor blocker; ASCVD = atherosclerotic cardiovascular disease; BMI = body mass index; BP = blood pressure; DBP = diastolic blood pressure; eGFR = estimated glomerular filtration rate; HDL = high-density lipoprotein; LVH = left ventricular hypertrophy; Q = quartile; SBP = systolic blood pressure; TC = total cholesterol; UACR = urine albumin-to-creatinine ratio.
Variability of eGFR and incident heart failure
Over a median follow-up period of 3.0 years (Q1-Q3: 2.2-3.7), 193 participants developed incident HF events (incidence rate per 1,000 person-years: 8.5 [95% CI: 7.4-9.8]) (Central Illustration). In unadjusted analyses, the cumulative incidence for incident HF was higher among participants with higher eGFR variability (Figure 1). The incidence of HF increased progressively across quartiles of eGFR variability. Participants in the lowest quartile (Q1) had an HF incidence rate of 4.8 per 1,000 person-years (95% CI: 3.3-6.9), while those in the highest quartile (Q4) had a rate of 12.6 per 1,000 person-years (95% CI: 10.0-15.9).
Central Illustration.
Variability in Estimated Glomerular Filtration Rate and Incidence of Heart Failure Among Individuals With Type 2 Diabetes
ACCORD = Action to Control Cardiovascular Risk in Diabetes; ACEI = angiotensin-converting enzyme inhibitor; ARB = angiotensin receptor blocker; CV = coefficient of variation; eGFR = estimated glomerular filtration rate.
Figure 1.
Cumulative Incidence of Heart Failure by Quartiles of eGFR Coefficient of Variation
eGFR = estimated glomerular filtration rate; HF = heart failure; Q = quartile.
In adjusted Cox regression models, the HRs for HF were significantly elevated in quartile 4 (Q4) compared to quartile 1 (Q1) across all models. In the fully adjusted model (model 5, Table 2), which accounted for demographic factors, clinical variables, and mean eGFR, the HR for HF in Q4 of eGFR CV was 1.88 (95% CI: 1.09-3.22, Ptrend = 0.019) compared to Q1. Each one-point increase in eGFR CV was associated with a 15% higher risk of HF (HR: 1.15; 95% CI: 0.99-1.33).
Table 2.
Rates and Hazard Ratios for Incident Heart Failure by eGFR Variability
| Quartiles of eGFR Coefficient of Variation |
Ptrend | Per 1-SD Increment | ||||
|---|---|---|---|---|---|---|
| Q1 (<6.23) | Q2 (6.23-9.77) | Q3 (9.78-14.57) | Q4 (>14.57) | |||
| No. of events/No. at risk | 27/1,901 | 41/1,900 | 53/1,900 | 72/1,900 | - | 193/7,601 |
| Person-years | 5,668.8 | 5,653.8 | 5,644.2 | 5,703.9 | - | 22,670.7 |
| Rate/1,000 person-years | 4.8 (3.3-6.9) | 7.3 (5.3-9.8) | 9.4 (7.2-12.3) | 12.6 (10.0-15.9) | - | 8.5 (7.4-9.8) |
| Model 1 | 1 (reference) | 1.41 (0.87-2.30) | 1.79 (1.12-2.85) | 2.32 (1.48-3.62) | <0.001 | 1.29 (1.14-1.44) |
| Model 2 | 1 (reference) | 1.33 (0.81-2.20) | 1.50 (0.92-2.44) | 1.87 (1.17-2.99) | 0.006 | 1.17 (1.03-1.32) |
| Model 3 | 1 (reference) | 1.32 (0.80-2.18) | 1.50 (0.92-2.44) | 1.85 (1.15-2.96) | 0.007 | 1.16 (1.02-1.32) |
| Model 4 | 1 (reference) | 1.3 (0.80-2.18) | 1.48 (0.91-2.42) | 1.85 (1.16-2.97) | 0.007 | 1.16 (1.02-1.31) |
| Model 5 | 1 (reference) | 1.32 (0.79-2.21) | 1.49 (0.89-2.52) | 1.88 (1.09-3.22) | 0.019 | 1.15 (0.99-1.33) |
Data are HRs (95% CI) unless otherwise specified. Model 1 adjusted for age, sex, race, and treatment arm; model 2 includes model 1 plus duration of diabetes, cigarette smoking, alcohol intake, mean body mass index, mean systolic blood pressure, mean total/high-density lipoprotein cholesterol ratio, mean glycated hemoglobin, mean urinary albumin-creatinine ratio, left ventricular hypertrophy, use of insulin/sulfonylurea, use of thiazolidinediones, use of antihypertensive medication, use of ACEI/ARB, use of beta blocker, and history of prevalent ASCVD; model 3 includes model 2 plus cardiac autonomic neuropathy; model 4 includes model 3 plus incident CAD as a time varying covariate; model 5 includes model 4 plus mean eGFR.
ACEI = angiotensin-converting enzyme inhibitor; ARB = angiotensin II receptor blocker; ASCVD = atherosclerotic cardiovascular disease; CAD = coronary artery disease; eGFR = estimated glomerular filtration rate; Q = quartile.
When examining incident HF risk using other variability metrics for eGFR, such as SD and VIM, the results were consistent with those observed using the CV. Higher quartiles of eGFR SD and VIM were associated with elevated rates of incident HF, with rates per 1,000 person-years increasing across quartiles. For eGFR SD, Q4 exhibited an adjusted HR of 1.44 (95% CI: 0.94-2.12) in the fully adjusted model. Similarly, for eGFR VIM, Q4 had an adjusted HR of 1.41 (95% CI: 0.94-2.12). In models assessing per 1-SD increment in eGFR variability, both SD and VIM demonstrated significant associations with HF risk, with HRs of 1.16 (95% CI: 1.01-1.33) and 1.18 (95% CI: 1.03-1.35, model 7), respectively (Supplemental Table 1).
In additional analyses, further adjustments for diuretic use (loop, thiazide, and potassium-sparing diuretics) and baseline prevalent retinopathy (models 6 and 7, respectively) did not materially alter the association between eGFR variability and incident HF. Compared to the reference quartile, the highest quartile of eGFR variability remained significantly associated with an increased risk of HF (Supplemental Tables 1 and 2).
We found no significant interaction between any of the eGFR variability measures and the treatment arms (including BP, lipid, and glycemic control treatment arms—all P for interaction >0.05).
Discussion
In this study, we found that greater variability in eGFR was significantly associated with a higher risk of incident HF among adults with type 2 diabetes. This association persisted across multiple metrics of eGFR variability, including the CV, SD, and VIM and was robust to extensive adjustment for potential confounders including demographic factors, clinical characteristics, baseline comorbidities, mean eGFR, and medication use. Sensitivity analyses incorporating additional adjustments for diuretic use and baseline retinopathy further confirmed the robustness of these findings. The strength and consistency of the observed associations across eGFR variability suggest that eGFR variability could be a clinically relevant marker of HF risk in this population, independent of traditional cardiovascular and renal risk factors.
Our study expands prior findings on the associations between eGFR and UACR and the incidence of HF.4, 5, 6, 7, 8 Most of the existing studies of kidney function and incident HF mainly assessed eGFR at a single time point and thus could not capture the extent of variability in kidney function over time. However, there are prior studies that assessed the rate of kidney function decline and the incidence of HF,15, 16, 17 which corroborate our findings, though we focused on overall variability of kidney function and not just decline, which is important given that hyperfiltration is an early feature of diabetic kidney disease.18 Our results are consistent with previous reports on a positive association between higher visit-to-visit variability in eGFR and increased risk of adverse CVD outcomes and mortality.10,19, 20, 21 However, these studies primarily focused on nondiabetic populations.10,19, 20, 21 Our study uniquely focuses on HF risk among individuals with T2DM and the examination of risk associations prospectively within a racially/ethnically diverse population.
Several mechanisms may explain the relation between eGFR variability and increased HF risk in adults with T2DM. eGFR variability may result from a number of physiologic and environmental influences, such as comorbid conditions, changes in renal plasma flow because of initiation or titration of many various medications, instances of acute kidney injury, fluctuation in fluid status, and age-related changes in creatinine levels, among others. eGFR variability may also reflect alteration in kidney regulatory mechanisms because of reduced renal reserve or nephron mass or vascular disease within renal arteries.22,23 A greater variability in eGFR can be viewed as a dynamic surrogate marker of reduced kidney resilience, signaling a diminished renal reserve and reflecting a limited ability to compensate in the setting of otherwise routine events that require the kidney to autoregulate.22,23 Consistent with this, prior studies have shown that greater eGFR variability associates with new or worsening kidney disease and higher risk of dialysis,24 which can all contribute to HF.15,16 Increased eGFR variability may represent not only a deficit in intrinsic kidney function but also the function of other organ systems involved in maintaining renal stability in the presence of external renal insults.20 eGFR variability may represent overall organ system resilience and adaptability. In this sense, increased eGFR variability could reflect a deficiency in myocardial function. Moreover, fluctuations in eGFR may reflect underlying hemodynamic instability and fibrosis, which promote left ventricular remodeling and impair cardiac compliance.9,25,26 Additionally, eGFR variability indicates episodic activation of inflammation, oxidative stress, or the RAAS, all of which are implicated in both kidney damage and myocardial dysfunction.9,25,26 Furthermore, eGFR variability could be linked to cardiac autonomic dysregulation, which is highly prevalent in T2DM and associated with both renal and cardiac dysfunction.27, 28, 29
Our findings have potential implications for clinical practice and future research. The possible clinical significance of eGFR variability includes 3 considerations. The assessment of the variability of eGFR, a routinely monitored laboratory value, can help improve HF risk stratification of patients with diabetes. Serial laboratory data are readily available in electronic health records during clinical visits, and such measures are already frequently obtained in the diabetes population. Increased eGFR variability can also be used as a target for treatment, aiming at improved outcome possibly without generating additional costs. Our results may have implications in terms of an increase in the frequency of monitoring kidney function among high-risk individuals with T2DM. However, prior to doing this, it is critical to ensure that well-known renoprotective medications shown to improve kidney function and also reduce HF hospitalizations (ACEi, ARB, sodium-glucose cotransporter-2 inhibitors, glucagon-like peptide-1 receptor agonists, and mineralocorticoid receptor antagonists30, 31, 32, 33), have the potential to reduce kidney function variability. Our results also point to the utmost importance of adherence to renoprotective medications to minimize kidney function variability and possibly reduce the risk of HF. From a research perspective, our study highlights the potential value of using eGFR variability as an outcome in clinical trials evaluating cardiorenal protective therapies to elucidate its role in predicting therapeutic benefits or risks.
Our study has several strengths, including the use of a large and diverse prospective cohort (including White, Black, and Hispanic individuals), the assessment of eGFR at regular intervals for all participants, the assessment of study outcomes using standardized protocols, the adjustment for mean eGFR and UACR over the study period, and a relatively long duration of follow-up. Moreover, we used more than 1 approach to assess eGFR variability, which minimizes the potential bias related to any of the approaches.34
Study Limitations
Our findings have a number of limitations. First, there is a possibility of unmeasured residual confounding given the observational design of our study. Second, the ACCORD study had a relatively short follow-up period (∼3 years), and a longer follow-up would have most probably led to an accrual of more HF events. Moreover, the ACCORD study did not collect data on left ventricular ejection fraction or HF subtype, precluding an investigation of the risks of HF with reduced ejection fraction and HF with preserved ejection fraction. Likewise, we did not have data on all renoprotective therapies and the extent to which participants adhered to these. Our estimation of the eGFR did not include markers of kidney function, such as serum cystatin C, as this may provide additional insight into the eGFR variability and HF associations. Finally, our study relied on only 7 time points to assess variability; thus, we possibly underestimated the variability and the magnitude of our effect estimates. Indeed, prior studies have shown that visit-to-visit variability of a biological variable increases with the number of visits used to calculate it.35
Conclusions
Our study demonstrates that higher eGFR variability was independently associated with an increased risk of incident HF in adults with T2DM. These findings highlight the potential clinical utility of monitoring eGFR variability as an indicator of HF risk in this high-risk population. Future research should aim to uncover the mechanisms driving this association and investigate whether interventions targeting eGFR variability can improve cardiovascular and renal outcomes as part of an integrated approach to risk management in diabetes care.
Perspectives.
COMPETENCY IN MEDICAL KNOWLEDGE: Variability in eGFR has been proposed as a novel risk factor for adverse cardiovascular outcomes. Its association with incident HF in adults with type 2 diabetes remains unclear.
COMPETENCY IN PATIENT CARE: Higher variability in eGFR is independently and robustly associated with an increased risk of incident HF in adults with type 2 diabetes. Hence, targeting individuals with high eGFR variability may help prevent HF.
TRANSLATIONAL OUTLOOK: eGFR variability can help refine the assessment of the risk of HF and provide new treatment targets for cardiorenal protection among individuals with type 2 diabetes. This could help target high-risk individuals who would benefit the most from cardioprotective therapies.
Funding support and author disclosures
Dr Echouffo Tcheugui was supported by NIH/NHLBI grant K23 HL153774. Dr Fonarow has consulted for Abbott, Amgen, AstraZeneca, Bayer, Boehringer Ingelheim, Cytokinetics, Eli Lilly, Johnson & Johnson, Medtronic, Merck, Novartis, and Pfizer. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Footnotes
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the Author Center.
Appendix
For supplemental tables and a figure, please see the online version of this paper.
Supplementary data
References
- 1.Dei Cas A., Khan S.S., Butler J., et al. Impact of diabetes on epidemiology, treatment, and outcomes of patients with heart failure. JACC Heart Fail. 2015;3:136–145. doi: 10.1016/j.jchf.2014.08.004. [DOI] [PubMed] [Google Scholar]
- 2.Aune D., Schlesinger S., Neuenschwander M., et al. Diabetes mellitus, blood glucose and the risk of heart failure: a systematic review and meta-analysis of prospective studies. Nutr Metab Cardiovasc Dis. 2018;28:1081–1091. doi: 10.1016/j.numecd.2018.07.005. [DOI] [PubMed] [Google Scholar]
- 3.Oktay A.A., Paul T.K., Koch C.A., et al. In: Endotext [Internet]. South Dartmouth (MA): MDText.com, Inc. Feingold K.R., Ahmed S.F., Anawalt B., et al., editors. 2000. Diabetes, Cardiomyopathy, and Heart Failure. [Updated 2023 Sep 26]https://www.ncbi.nlm.nih.gov/books/NBK560257/ [Google Scholar]
- 4.Kottgen A., Russell S.D., Loehr L.R., et al. Reduced kidney function as a risk factor for incident heart failure: the atherosclerosis risk in communities (ARIC) study. J Am Soc Nephrol. 2007;18:1307–1315. doi: 10.1681/ASN.2006101159. [DOI] [PubMed] [Google Scholar]
- 5.Nichols G.A., Gullion C.M., Koro C.E., Ephross S.A., Brown J.B. The incidence of congestive heart failure in type 2 diabetes: an update. Diabetes Care. 2004;27:1879–1884. doi: 10.2337/diacare.27.8.1879. [DOI] [PubMed] [Google Scholar]
- 6.Vaur L., Gueret P., Lievre M., Chabaud S., Passa P. Development of congestive heart failure in type 2 diabetic patients with microalbuminuria of proteinuria: observations from the DIABHYCAR (type 2 DIABetes, Hypertension, CArdiovascular Events and Ramipril) study. Diabetes Care. 2003;26:2489. doi: 10.2337/diacare.26.3.855. [DOI] [PubMed] [Google Scholar]
- 7.Tancredi M., Rosengren A., Olsson M., et al. The relationship between three eGFR formulas and hospitalization for heart failure in 54 486 individuals with type 2 diabetes. Diabetes Metab Res Rev. 2016;32:730–735. doi: 10.1002/dmrr.2793. [DOI] [PubMed] [Google Scholar]
- 8.Vestberg D., Rosengren A., Olsson M., et al. Decreased eGFR as a risk factor for heart failure in 13 781 individuals with type 1 diabetes. J Diabetes Sci Technol. 2015;10:131–136. doi: 10.1177/1932296815596174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Zannad F., Rossignol P. Cardiorenal syndrome revisited. Circulation. 2018;138:929–944. doi: 10.1161/CIRCULATIONAHA.117.028814. [DOI] [PubMed] [Google Scholar]
- 10.Malhotra R., Katz R., Jotwani V., et al. Estimated GFR variability and risk of cardiovascular events and mortality in SPRINT (systolic blood pressure intervention trial) Am J kidney Dis. 2021;78:48–56. doi: 10.1053/j.ajkd.2020.10.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Buse J.B., Bigger J.T., Byington R.P., et al. Action to control cardiovascular risk in diabetes (ACCORD) trial: design and methods. Am J Cardiol. 2007;99:21i–33i. doi: 10.1016/j.amjcard.2007.03.003. [DOI] [PubMed] [Google Scholar]
- 12.Mychaleckyj J.C., Craven T., Nayak U., et al. Reversibility of fenofibrate therapy-induced renal function impairment in ACCORD type 2 diabetic participants. Diabetes Care. 2012;35:1008–1014. doi: 10.2337/dc11-1811. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ismail-Beigi F., Craven T., Banerji M.A., et al. Effect of intensive treatment of hyperglycaemia on microvascular outcomes in type 2 diabetes: an analysis of the ACCORD randomised trial. Lancet. 2010;376:419–430. doi: 10.1016/S0140-6736(10)60576-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Levey A.S., Stevens L.A., Schmid C.H., et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150:604–612. doi: 10.7326/0003-4819-150-9-200905050-00006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Sandholm N., Valo E., Tuomilehto J., Harjutsalo V., Groop P.-H. Rate of kidney function decline is associated with kidney and heart failure in individuals with type 1 diabetes. Kidney Int reports. 2023;8:2043–2055. doi: 10.1016/j.ekir.2023.07.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Bueno Junior C.R., Bano A., Tang Y., et al. Rapid kidney function decline and increased risk of heart failure in patients with type 2 diabetes: findings from the ACCORD cohort: rapid kidney function decline and heart failure in T2D. Cardiovasc Diabetol. 2023;22:131. doi: 10.1186/s12933-023-01869-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Ragot S., Saulnier P.-J., Velho G., et al. Dynamic changes in renal function are associated with major cardiovascular events in patients with type 2 diabetes. Diabetes Care. 2016;39:1259–1266. doi: 10.2337/dc15-2607. [DOI] [PubMed] [Google Scholar]
- 18.Mohandes S., Doke T., Hu H., Mukhi D., Dhillon P., Susztak K. Molecular pathways that drive diabetic kidney disease. J Clin Invest. 2023;133 doi: 10.1172/JCI165654. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Jun M., Harris K., Heerspink H.J.L., et al. Variability in estimated glomerular filtration rate and the risk of major clinical outcomes in diabetes: post hoc analysis from the ADVANCE trial. Diabetes Obes Metab. 2021;23:1420–1425. doi: 10.1111/dom.14351. [DOI] [PubMed] [Google Scholar]
- 20.Al-Aly Z., Balasubramanian S., McDonald J.R., Scherrer J.F., O’Hare A.M. Greater variability in kidney function is associated with an increased risk of death. Kidney Int. 2012;82:1208–1214. doi: 10.1038/ki.2012.276. [DOI] [PubMed] [Google Scholar]
- 21.Suzuki A., Obi Y., Hayashi T., et al. Visit-to-visit variability in estimated glomerular filtration rate predicts hospitalization and death due to cardiovascular events. Clin Exp Nephrol. 2019;23:661–668. doi: 10.1007/s10157-019-01695-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Malhotra R., Chopra T., Ix J.H. Interpreting estimated GFR variability and its clinical significance. Kidney Med. 2023;5 doi: 10.1016/j.xkme.2023.100599. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Jufar A.H., Lankadeva Y.R., May C.N., Cochrane A.D., Bellomo R., Evans R.G. Renal functional reserve: from physiological phenomenon to clinical biomarker and beyond. Am J Physiol Regul Integr Comp Physiol. 2020;319:R690–R702. doi: 10.1152/ajpregu.00237.2020. [DOI] [PubMed] [Google Scholar]
- 24.Tseng C.-L., Lafrance J.-P., Lu S.-E., et al. Variability in estimated glomerular filtration rate values is a risk factor in chronic kidney disease progression among patients with diabetes. BMC Nephrol. 2015;16:34. doi: 10.1186/s12882-015-0025-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Rockey D.C., Bell P.D., Hill J.A. Fibrosis--a common pathway to organ injury and failure. N Engl J Med. 2015;372:1138–1149. doi: 10.1056/NEJMra1300575. [DOI] [PubMed] [Google Scholar]
- 26.Travers J.G., Kamal F.A., Robbins J., Yutzey K.E., Blaxall B.C. Cardiac fibrosis: the fibroblast awakens. Circ Res. 2016;118:1021–1040. doi: 10.1161/CIRCRESAHA.115.306565. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Kaze A.D., Yuyun M.F., Erqou S., Fonarow G.C., Echouffo-Tcheugui J.B. Cardiac autonomic neuropathy and risk of incident heart failure among adults with type 2 diabetes. Eur J Heart Fail. 2022;24:634–641. doi: 10.1002/ejhf.2432. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Tang Y., Ang L., Jaiswal M., et al. Cardiovascular autonomic neuropathy and risk of kidney function decline in type 1 and type 2 diabetes: findings from the PERL and ACCORD cohorts. Diabetes. 2024;73:751–762. doi: 10.2337/db23-0247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Cho I., Lim S., Kwon M., et al. Cardiovascular autonomic neuropathy and the risk of diabetic kidney disease. Front Endocrinol (Lausanne) 2024;15 doi: 10.3389/fendo.2024.1462610. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Neuen B.L., Fletcher R.A., Heath L., et al. Cardiovascular, kidney and safety outcomes with GLP-1 receptor agonists alone and in combination with SGLT2 inhibitors in type 2 diabetes: a systematic review and meta-analysis. Circulation. 2024;150:1781–1790. doi: 10.1161/CIRCULATIONAHA.124.071689. [DOI] [PubMed] [Google Scholar]
- 31.Perkovic V., Tuttle K.R., Rossing P., et al. Effects of semaglutide on chronic kidney disease in patients with type 2 diabetes. N Engl J Med. 2024;391:109–121. doi: 10.1056/NEJMoa2403347. [DOI] [PubMed] [Google Scholar]
- 32.Bakris G.L., Agarwal R., Anker S.D., et al. Effect of finerenone on chronic kidney disease outcomes in type 2 diabetes. N Engl J Med. 2020;383:2219–2229. doi: 10.1056/NEJMoa2025845. [DOI] [PubMed] [Google Scholar]
- 33.Filippatos G., Anker S.D., Agarwal R., et al. Finerenone reduces risk of incident heart failure in patients with chronic kidney disease and type 2 diabetes: analyses from the FIGARO-DKD trial. Circulation. 2022;145:437–447. doi: 10.1161/CIRCULATIONAHA.121.057983. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Barrett J.K., Huille R., Parker R., Yano Y., Griswold M. Estimating the association between blood pressure variability and cardiovascular disease: an application using the ARIC Study. Stat Med. 2019;38:1855–1868. doi: 10.1002/sim.8074. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Levitan E.B., Kaciroti N., Oparil S., Julius S., Muntner P. Blood pressure measurement device, number and timing of visits, and intra-individual visit-to-visit variability of blood pressure. J Clin Hypertens (Greenwich) 2012;14:744–750. doi: 10.1111/jch.12005. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.