Author + information
- Received September 13, 2013
- Revision received November 13, 2013
- Accepted November 21, 2013
- Published online May 1, 2014.
- Geoffrey M. Crimmins, MD,
- Ryan D. Madder, MD,
- Victor Marinescu, MD and
- Robert D. Safian, MD∗ ()
- Department of Cardiovascular Medicine and Oakland University William Beaumont School of Medicine, Beaumont Health System, Royal Oak, Michigan
- ↵∗Reprint requests and correspondence:
Dr. Robert D. Safian, Center for Innovation and Research in Cardiovascular Disease (CIRC), Beaumont Health System, Oakland University William Beaumont School of Medicine, Royal Oak, Michigan 48073.
Objectives The purpose of this study was to evaluate the validity of estimates of glomerular filtration rate (eGFR) for assessing serial changes in renal function after renal artery stenting.
Background eGFR are unreliable for assessing serial renal function in patients with atherosclerotic renal artery stenosis (RAS). eGFR have not been validated for assessment of serial renal function after renal artery stenting.
Methods Serum creatinine (SCr) and 125I-iothalamate GFR (iGFR) were measured in RAS patients before and after renal artery stenting. eGFR were calculated from Modification of Diet in Renal Disease (MDRD), Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI), and Cockcroft-Gault (CG) formulas. Using iGFR as the reference standard, the sensitivity, specificity, and area under the receiver-operating characteristic curve (AUC) were determined for MDRD, CKD-EPI, and CG for assessing changes in GFR before and after intervention.
Results Between 1998 and 2007, 84 patients underwent iGFR and eGFR before and after renal artery stenting. All eGFR demonstrated poor sensitivity and reliability for detecting ≥20% changes in iGFR, and poor agreement in the magnitude and direction of change in iGFR, before and after renal stenting.
Conclusions In RAS patients, eGFR demonstrate poor sensitivity and reliability for detecting meaningful changes in iGFR after renal artery stenting. eGFR should be abandoned as primary endpoints in major clinical trials assessing the impact of renal revascularization on renal function.
Atherosclerotic renal artery stenosis (RAS) is highly prevalent among elderly patients with hypertension, chronic kidney disease (CKD), and coronary and peripheral arterial disease. Although renal artery revascularization is most often performed by stenting, several recent randomized trials reported no improvement in renal function with stenting compared with medical therapy (1–3), based on estimates of glomerular filtration rates (eGFR) to assess serial renal function. However, a recent study demonstrated that compared with measured glomerular filtration rate (GFR), eGFR have poor sensitivity and reliability for detecting 20% changes in measured GFR, and recommended that eGFR not be used to assess serial GFR in patients with RAS (4). The purpose of this study is to evaluate the validity of eGFR for assessing serial changes in renal function after renal artery stenting.
Patient selection and GFR assessment
Between 1998 and 2007, serial 125I-iothalamate glomerular filtration rates (iGFR) were measured in 254 patients with RAS, as previously described (4). Within this population, 81 patients underwent renal artery stenting and are the focus of the present study. Criteria for renal revascularization were previously defined (5). Briefly, these criteria included those patients with unilateral or bilateral RAS ≥70% and clinical evidence for hypertensive crisis associated with nonischemic pulmonary edema or acute neurological injury; patients with severe hypertension alone, known renal parenchymal disease, proteinuria ≥1.0 g in 24 h, or anticipated life expectancy ≤2 years were excluded. GFR was measured by the plasma disappearance of iGFR using a 2-compartment pharmacokinetic model (6), as detailed previously (4). Briefly, blood samples were drawn 5, 10, and 15 min after administration of 0.15 ml of 125I-iothalamate, and at 30-min intervals starting 3 h after administration. eGFR were calculated according to the 4-variable Modification of Diet in Renal Disease (MDRD), Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI), and Cockcroft-Gault (CG) formulas (7–9), as previously described (4). The percent changes in iGFR and eGFR were calculated for all patients after renal stenting. For comparison of iGFR and eGFR, patients were included if iGFR were available before and after renal stenting, and patients were on stable doses of antihypertensive and diuretic medications between serum creatinine (SCr) and iGFR measurements. SCr was measured by the Beaumont Reference Laboratory using the modified Jaffé reaction indirectly traceable to isotope dilution mass spectrometry (Roche Modular Instruments, Roche Diagnostics, Indianapolis, Indiana) as previously described (4). This study was approved by the Human Investigations Committee of Beaumont Health System.
iGFR and eGFR were compared with each other before and after stenting using paired Student t tests, and the percent changes in iGFR after stenting were compared with percent changes in eGFR. Compared with baseline, iGFR after stenting was considered stable if the iGFR increased or decreased by <20%, increased if iGFR increased by ≥20%, and decreased if iGFR decreased by ≥20%. Pearson correlation coefficients were used to study the correlation between percent changes in iGFR and eGFR after stenting. Because a 20% change in eGFR has been used as a renal endpoint in randomized trials of renal revascularization (10,11), a κ-statistic was generated to assess the degree of agreement between 20% changes in iGFR and eGFR. The sensitivity, specificity, and positive and negative predictive values of eGFR were determined for identifying a ≥20% increase and a ≥20% decrease in iGFR after stenting. Receiver-operating characteristic curves were constructed, and the area under the curve (AUC) was calculated for identifying a ≥20% increase or decrease in iGFR. The reliability of the estimates was characterized as excellent (AUC ≥0.9), good (AUC = 0.8 to 0.89), fair (AUC = 0.7 to 0.79), and poor (AUC < 0.7). Categorical variables are reported as frequencies. Continuous variables, including SCr, age, and time, are reported as mean ± SD, and all others are reported as median (25th to 75th percentiles). Bland-Altman analysis was performed using Microsoft Excel Analyze-IT 210 (Microsoft Corporation, Redmond, Washington); other analyses were performed using SAS for Windows (version 9.2, Cary, North Carolina).
Between 1998 and 2007, 81 RAS patients underwent renal artery stenting and had iGFR and SCr measured before and after renal stenting. Three patients had repeat intervention more than 1 year following initial intervention and were considered as separate entries, producing a total of 84 patients (Table 1). GFR was performed 87 ± 122 days before stenting and 180 ± 171 days after stenting. SCr was performed within 2 weeks of iGFR in 81% of patients.
Baseline renal function
Before revascularization, the measured iGFR was 54.0 ± 19.9 ml/min/1.73 m2. Baseline eGFR were 51.4 ± 20.2 ml/min/1.73 m2 using MDRD, 49.9 ± 19.8 ml/min/1.73 m2 using CKD-EPI, and 52.1 ± 18.4 ml/min/1.73 m2 using CG. At baseline, iGFR was <60 ml/min/1.73 m2 in 58 (69.1%) measurements, consistent with CKD. At baseline, eGFR was <60 ml/min/1.73 m2 in 65.5% by MDRD, 67.9% by CKD-EPI, and 66.7% by CG. Baseline eGFR incorrectly diagnosed CKD in 2.4% to 8.3% patients and missed CKD in 9.5% to 11.9%.
Change in renal function after stenting
iGFR before (54.0 ± 19.9 ml/min/1.73 m2) and after (56.2 ± 24.3 ml/min/1.73 m2) stenting were similar (Table 2). After revascularization, iGFR increased in 22.6%, decreased in 14.3%, and remained stable in 63.1% of patients.
There was no difference in eGFR before and after stenting for MDRD (51.4 ± 20.2 ml/min/1.73 m2 vs. 51.6 ± 23.7 ml/min/1.73 m2, p = NS), CKD-EPI (49.4 ± 19.8 ml/min/1.73 m2 vs. 48.9 ± 21.3 ml/min/1.73 m2, p = NS), or CG (63.2 ± 20.7 ml/min/1.73 m2 vs. 64.2 ± 13.4 ml/min/1.73 m2, p = NS) (Table 2). Whereas iGFR identified any increase in GFR after revascularization in 57.1% of patients, only 40.5% of patients had any increase in eGFR by MDRD, 40.5% by CKD-EPI, and 53.9% by CG. Correlations between percent changes in iGFR and eGFR were poor: MDRD r2 = 0.32 (p = 0.0034); CKD-EPI r2 = 0.31 (p= 0.0037); CG r2 = 0.32 (p = 0.05) (Fig. 1). The median differences (biases) between the estimates and iGFR were 3.7 ± 3.2, 4.4 ± 3.2, and 3.8 ± 2.7 ml/min/1.73 m2 for MDRD, CKD-EPI, and CG, respectively. The precision of GFR estimates demonstrated interquartile ranges of −11.2 to 19.1, −10.8 to 19.6, and −11.0 to 21.2 ml/min/1.73 m2 for MDRD, CKD-EPI, and CG, respectively (Fig. 2). After revascularization, the direction of change in eGFR was discordant with the direction of change in iGFR in 38.1% of MDRD, 38.1% of CKD-EPI, and 46.2% of CG.
A ≥20% increase was identified by iGFR in 22.6% of patients, by MDRD in 16.7%, by CKD-EPI in 16.7%, and by CG in 10.3% (Table 3). The degree of agreement to detect ≥20% increase in iGFR was poor for MDRD (κ = 0.29, 95% confidence interval [CI]: 0.05 to 0.53), CKD-EPI (κ = 0.29, 95% CI: 0.05 to 0.53), and CG (κ = 0.23, 95% CI: −0.14 to 0.59). A ≥20% decrease was identified by iGFR in 14.3% of patients, by MDRD in 17.9%, by CKD-EPI in 19.1%, and by CG in 15.4% (Table 3). These data suggest that serial eGFR consistently underestimate the frequency of improvement in iGFR, and overestimate the frequency of decline in iGFR. The degree of agreement to detect ≥20% decrease in iGFR was poor for MDRD (κ = 0.52, 95% CI: 0.27 to 0.76), CKD-EPI (κ = 0.49, 95% CI: 0.24 to 0.74), and CG (κ = 0.26, 95% CI: −014. to 0.66).
Overall, GFR estimates were characterized by poor sensitivity, modest specificity, and poor reliability for identifying ≥20% increase (AUC: 0.75 to 0.77) and ≥20% decrease (AUC: 0.69 to 0.72) in iGFR (Table 4, Fig. 3).
Epidemiological studies suggest a strong association between cardiovascular mortality and stages of CKD when classified by eGFR (12). Use of eGFR has also led to recognition of earlier stages of CKD in elderly patients, and may result in less drug toxicity when drug dosages are adjusted for eGFR. However, eGFR are less reliable at higher levels of measured GFR (13), and may misclassify the stage of CKD in 35% of patients (14). Although eGFR is usually calculated using equations based on SCr, several studies suggest that eGFR equations using serum cystatin-C (alone or in combination with SCr) may be more accurate (15,16). However, eGFR equations using cystatin-C have not been validated for assessment of serial GFR, and cystatin-C was not measured in our study.
Patients with RAS may have a combination of intrinsic kidney disease and reduction in GFR and renal blood flow (17). Randomized trials of medical therapy versus renal artery stenting failed to demonstrate any benefit of renal revascularization, frequently relying on ≥20% changes in eGFR as the primary renal endpoint (10,11). However, a more recent study of RAS patients demonstrated that eGFR are not valid for assessment of serial renal function (4). In that study, the CKD stage was incorrectly classified by eGFR in nearly one-third of patients; there were poor correlations between 20% changes in eGFR and iGFR; and the direction of change (increase or decrease in GFR) was discordant in up to 40% of patients.
The present study extends the findings of our previous study (4) to the subset of RAS patients who undergo renal artery stenting. eGFR are unreliable for detecting ≥20% changes in iGFR and for identifying the direction and magnitude of change in iGFR after renal stenting. Furthermore, compared with iGFR, eGFR underestimates improvement and overestimates deterioration in serial renal function after stenting. Compared with baseline, the percent changes in iGFR and eGFR and the degree of agreement between iGFR and eGFR after stenting are poor.
The present study was designed to evaluate the validity of eGFR for assessing changes in measured GFR after renal stenting. The study was not designed to compare medical therapy with revascularization; assess important cardiovascular endpoints such as death, stroke, myocardial infarction, and heart failure; or evaluate the impact of renal revascularization on blood pressure or renal function. Other measures of eGFR were not evaluated because the frequency was too low (doubling of SCr) or data were not collected (serum cystatin-C).
In patients with RAS, eGFR are unreliable for detecting ≥20% changes in measured GFR after renal stenting. eGFR are invalid endpoints for assessing serial GFR in patients with RAS, and should be abandoned as major endpoints in trials of renal revascularization.
The authors thank Judy Boura, MS, for her expertise with statistical analysis.
Dr. Madder has received research support and speaker honoraria from Infraredx; and consulting fees from St. Jude Medical. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- area under the receiver-operating characteristic curve
- confidence interval
- chronic kidney disease
- Chronic Kidney Disease Epidemiology Collaboration
- estimated glomerular filtration rate
- 125I-iothalamate glomerular filtration rate
- Modification of Diet in Renal Disease
- atherosclerotic renal artery stenosis
- serum creatinine
- Received September 13, 2013.
- Revision received November 13, 2013.
- Accepted November 21, 2013.
- American College of Cardiology Foundation
- Madder R.D.,
- Hickman L.,
- Crimmins G.M.,
- et al.
- Matsushita K.,
- van der Velde M.,
- Astor B.C.,
- et al.,
- Chronic Kidney Disease Prognosis Consortium
- Rule A.D.
- Botev R.,
- Mallié J.P.,
- Wetzels J.F.M.,
- Couchoud C.,
- Schück O.
- Safian R.D.,
- Madder R.D.