Abstract
Background and objectives CKD is associated with increased cardiovascular risk not fully attributable to traditional risk factors. We compared endothelium-dependent and -independent vascular function among individuals with advanced CKD with function in those with vascular disease but preserved kidney function.
Design, setting, participants, & measurements Matched cohort analysis randomly selected from 1259 participants at a single center with measurements of brachial artery flow–mediated dilation, an endothelium-dependent process, and nitroglycerin-mediated dilation, an endothelium-independent process. Patients with advanced CKD (n=70) were matched 1:1 to controls with preserved kidney function and (1) no overt vascular disease, (2) hypertension, and (3) coronary artery disease.
Results The trend toward lower flow-mediated dilation (mean±SEM) in advanced CKD (5.4%±0.5%) compared with no overt vascular disease (7.3%±0.6%), hypertension (6.2%±0.5%), and coronary artery disease (5.8%±0.5%) did not reach statistical significance in adjusted analyses (P=0.05). Nitroglycerin-mediated dilation was lower in advanced CKD compared with in the other groups (adjusted nitroglycerin-mediated dilation: 6.9%±0.8%, 11.8%±0.9%, 11.0%±0.7%, and 10.5%±0.7% in advanced CKD, no overt vascular disease, hypertension, and coronary artery disease groups, respectively; P<0.001). Using tertiles generated from the full cohort and no overt vascular disease as the reference, the adjusted odds of flow-mediated dilation falling within the lowest tertile was higher in both the advanced CKD (odds ratio, 4.84; 95% confidence interval, 2.09 to 11.25) and coronary artery disease (odds ratio, 4.17; 95% confidence interval, 1.76 to 9.87) groups. In contrast, the adjusted odds of lowest tertile nitroglycerin-mediated dilation was higher in advanced CKD (odds ratio, 24.25; 95% confidence interval, 7.16 to 82.13) but not in the hypertension (odds ratio, 0.79; 95% confidence interval, 0.23 to 2.77) or coronary artery disease (odds ratio, 2.34; 95% confidence interval, 0.74 to 7.40) group.
Conclusions Impairment in endothelium-dependent vascular function is present in patients with CKD and those with clinically evident vascular disease but preserved kidney function. In contrast, substantial reduction in endothelium-independent function was observed only in the CKD group, suggesting differences in severity and pathophysiology of vascular dysfunction between CKD and other disease states.
- arteries
- cardiovascular disease
- chronic kidney disease
- chronic renal disease
- chronic renal insufficiency
- end-stage renal disease
- endothelium
- vascular disease
- nitric oxide
- Accidental Falls
- Brachial Artery
- Cohort Studies
- coronary artery disease
- Dilatation
- Endothelium, Vascular
- Humans
- hypertension
- Nitroglycerin
- Renal Insufficiency, Chronic
- risk factors
Introduction
Patients with CKD have a high burden of cardiovascular morbidity and mortality compared with the general population, and this is not adequately predicted by traditional risk factors (1,2). Abnormalities in both endothelium-dependent and -independent processes have been linked to adverse outcomes in patients at high risk for cardiovascular events (3–6). Impaired endothelial function, as indicated by reduced brachial artery flow-mediated dilation (FMD), has been reported in CKD (7); however, previous studies have not directly compared FMD in patients with CKD with other populations affected by vascular disease (8–10). Additionally, the contribution of endothelium-independent processes to the observed reductions in FMD has not been addressed. Oxidant- or inflammation-mediated alterations in vascular smooth muscle cell function and structural changes, such as vascular calcification, are potential causes of endothelium-independent vascular dysfunction in CKD (11–13).
We hypothesized that impairment in vascular function is greater in patients with advanced CKD compared with those with clinically evident vascular disease but preserved kidney function and that both endothelium-dependent and -independent processes are affected. To test these hypotheses, we examined brachial artery FMD and nitroglycerin-mediated dilation (NMD) in patients with advanced CKD and three age- and sex-matched control groups of individuals with hypertension, coronary artery disease, or no overt vascular disease.
Materials and Methods
Participants
Four cohorts were assembled from 1259 participants in research studies of vascular function performed at Boston University School of Medicine between 1999 and 2009 (3,4,14–23). All participants provided informed consent under protocols approved by the Boston University Medical Center Institutional Review Board. The advanced CKD cohort included all participants in a prospective single-center study of predictors of hemodialysis arteriovenous (AV) fistula maturation. The study enrolled 70 patients between July of 2008 and November of 2009 who were scheduled to have creation of an AV fistula and were either receiving treatment with maintenance hemodialysis or expected to begin treatment with maintenance hemodialysis within 6 months. FMD and NMD were measured before AV fistula creation. To develop the non-CKD control cohorts, the remaining participants in the vascular function research studies were categorized as having no overt vascular disease, hypertension, or coronary artery disease on the basis of data collected at study enrollment through participant report and medical record review. No overt vascular disease was defined as no history of hypertension, coronary artery disease, or peripheral artery disease. Hypertension was defined as BP>140/90 mmHg or use of antihypertensive medications to treat hypertension. Coronary artery disease was defined as a history of myocardial infarction or evidence of coronary artery disease by angiography or stress testing. After categorization into vascular disease groups and exclusion of those with serum creatinine of 2.0 mg/dl or greater, 70 members of each control group were selected randomly using a random number list, with frequency matching to the patients with advanced CKD by sex and age using intervals of 10 years.
Brachial Artery Reactivity Measurement
Brachial artery reactivity was evaluated in a single laboratory for all participants using a standardized procedure as previously described (24,25). Subjects were placed in a supine position in a temperature-controlled room. A BP cuff was placed on the upper arm, and after 10 minutes of rest, digitized two-dimensional images of the brachial artery were obtained using a high-frequency ultrasound probe. FMD was determined by imaging the brachial artery before and 1 minute after occlusion of the brachial artery for 5 minutes. Occlusion of the brachial artery was achieved by inflating the BP cuff to 50 mmHg above the systolic BP. NMD was determined through imaging of the brachial artery before and 4 minutes after sublingual administration of nitroglycerin (0.4 mg). For the subjects with advanced CKD, the arm in which fistula creation was planned was used unless there was a preexisting AV vascular access in that arm. A period of at least 10 minutes after FMD assessment was required before measuring NMD to allow the brachial artery to return to baseline. Subjects did not undergo determination of NMD if there was known intolerance of nitroglycerin, a history of migraine headaches, systolic BP of <100 mmHg, or use of a phosphodiesterase-5 inhibitor during the previous week. Both FMD and NMD are expressed as percentage increases in diameter compared with baseline diameter. Specialized software (Brachial Tools; Medical Imaging Applications, Inc., Iowa City, IA) was used for determination of brachial artery diameter from digitized end diastolic images (R wave gated) transmitted from the ultrasound system (Cardiovascular Engineering Inc., Holliston, MA) (24).
Statistical Analyses
Demographic and clinical characteristics were compared with one-way ANOVA for continuous variables and the chi-squared test for categorical variables. Linear regression models were constructed with either FMD or NMD as the dependent variable as a function of the cohort group (no overt vascular disease, hypertension, coronary artery disease, or advanced CKD). Models were adjusted for age and sex. Models were additionally developed using a backward selection procedure with P<0.05 as the tolerance for keeping variables in the model. Prespecified covariates were baseline brachial artery diameter, race, peripheral arterial disease, diabetes, current smoking history, total cholesterol, and body mass index. Post hoc pairwise comparisons of between-cohort differences with respect to FMD and NMD with Bonferroni adjustment for multiplicity were conducted within the ANOVA and linear models.
In secondary analyses, multivariable logistic regression was used to determine the odds ratios (OR) for FMD or NMD falling within the lowest tertile. The tertiles were generated by combining the four groups, and the reference for the ORs was the no overt vascular disease group. To measure the relationship between kidney function and vascular function, the four predefined groups were combined and divided according to quartile of eGFR estimated using the Modification of Diet in Renal Disease Study equation (26). Linear regression models were constructed with either FMD or NMD as the dependent variable as a function of eGFR quartile. For both linear and logistic regression, covariates were included in adjusted models using the same criteria as for the primary analyses.
To assess potential effect modification of coronary artery disease and CKD, an interaction term between these two categories was created for general linear models for FMD and NMD. Models were adjusted for age and sex.
General linear models were used to analyze associations between clinical and biologic predictors and FMD and NMD in the advanced CKD cohort. Models were adjusted for age and sex.
An α level of <0.05 was used to define statistical significance for each pairwise comparison. Statistical analysis was performed using the Statistical Package for Social Sciences (SPSS 21.0, Chicago, IL).
Results
Participant Characteristics
Table 1 shows the clinical characteristics of the four cohorts. Of the 280 participants who had FMD measurement, 246 also had NMD measurement. Per design, age and sex distributions were similar across groups. Compared with the control groups, the advanced CKD cohort had more black participants and a greater proportion with diabetes. The proportion of participants treated with cardiovascular medications was highest in the advanced CKD and coronary artery disease groups. Sixty-six percent of the participants with advanced CKD were receiving RRT. For the participants in the advanced CKD group not receiving RRT, the median (interquartile range) eGFR was 12 (8–14) ml/min per 1.73 m2.
Characteristics of advanced CKD and control groups
Vascular Function
The trend toward lower FMD (mean±SEM) in advanced CKD (5.4%±0.5%) compared with no overt vascular disease (7.3%±0.6%), hypertension (6.2%±0.5%), or coronary artery disease (5.8%±0.5%) did not reach statistical significance in adjusted analyses (P=0.05) (Figure 1, Table 2). Mean NMD was lower for the advanced CKD group compared with the control groups in both the unadjusted and adjusted models (P=0.002 for the full multivariable model) (Figure 1, Table 2).
Mean reduction (percentage) in adjusted flow-mediated dilation (FMD) and nitroglycerin-mediated dilation (NMD) compared with no overt vascular disease group for subjects who had both measurements (n=246). Final multivariable model for FMD included age, sex, and baseline brachial artery diameter. Final multivariable model for NMD included age, sex, diabetes, peripheral artery disease, and baseline brachial artery diameter. CAD, coronary artery disease; HTN, hypertension. *P<0.05 compared with no overt vascular disease after Bonferroni adjustment for multiple comparisons.
Vascular function measurements for advanced CKD and control groups
Using the no overt vascular disease group as the reference, the adjusted OR for FMD in the lowest tertile was 4.84 (95% confidence interval [95% CI], 2.09 to 11.25) for the patients with advanced CKD. This was similar to the adjusted OR for the coronary artery disease group (adjusted OR, 4.17; 95% CI, 1.76 to 9.87) (Table 3). The risk for the lowest tertile NMD was higher for the advanced CKD group (adjusted OR, 24.25; 95% CI, 7.16 to 82.13) but was not higher for either the hypertension (adjusted OR, 0.79; 95% CI, 0.23 to 2.77) or the coronary artery disease (adjusted OR, 2.34; 95% CI, 0.74 to 7.40) group (Table 3). The presence of coronary artery disease did not modify the effect of CKD on vascular function (P value for interaction =0.84 for FMD; P value for interaction =0.29 for NMD).
Odds ratios for low flow-mediated dilation and nitroglycerin-mediated dilation compared with no overt vascular disease group
To further explore the relationship between kidney function and vascular function, we classified the entire sample by eGFR quartiles (Figure 2).
Multivariable-adjusted flow-mediated dilation (FMD), nitroglycerin-mediated dilation (NMD), and eGFR quartile. Values are expressed as least square mean (SEM). FMD: quartile 1: eGFR>109 ml/min per 1.73 m2; quartile 2: eGFR=84–109 ml/min per 1.73 m2; quartile 3: eGFR=34–83 ml/min per 1.73 m2; quartile 4: eGFR<34 ml/min per 1.73 m2. NMD: quartile 1: eGFR>110 ml/min per 1.73 m2; quartile 2: eGFR=86–110 ml/min per 1.73 m2; quartile 3: eGFR=54–85 ml/min per 1.73 m2; quartile 4: eGFR<54 ml/min per 1.73 m2. Multivariable model for FMD includes age, sex, diabetes, and baseline brachial artery diameter. Multivariable model for NMD includes age, sex, diabetes, baseline brachial artery diameter, and peripheral artery disease. *P<0.05 compared with quartile 4 after Bonferroni adjustment for multiple comparisons.
For both FMD and NMD, the relationship between vascular function and eGFR had an inverse J shape with progressively lower brachial artery dilation values in the second, third, and fourth eGFR quartiles. The lower FMD and NMD values in the first compared with second eGFR quartiles are similar to patterns observed for relationships between eGFR and other adverse outcomes and have been attributed to low muscle mass rather than better kidney function as the determinant of the highest eGFR values (27,28).
For the advanced CKD group, factors associated with poor vascular function in age- and sex-adjusted analyses include longer dialysis vintage for both FMD and NMD and treatment with dialysis, higher systolic BP, and higher diastolic BP for NMD (Table 4).
Associations between clinical factors and vascular function in CKD group
Discussion
In this matched cohort analysis, we found that (1) vascular function was impaired in the individuals with advanced CKD compared with those with no overt vascular disease and that (2) NMD was markedly reduced in advanced CKD but preserved in those with hypertension or coronary artery disease. Notably, in contrast to the CKD group, NMD was not affected in the coronary artery disease group, despite a reduction in FMD that was of similar magnitude to that observed in CKD. These findings suggest that vascular impairment in advanced CKD is more profound than in hypertension or coronary artery disease and that the pathophysiology of vascular dysfunction may differ between CKD and other disease states.
FMD measures the ability of the brachial artery to respond with endothelial nitric oxide release during reactive hyperemia. Because nitric oxide release requires a healthy endothelium, reduced FMD is considered an indicator of impaired endothelial function. However, in addition to a well functioning endothelium, the flow-mediated vasodilatory response requires arterial responsiveness to nitric oxide, a process that is endothelium independent. Impaired nitric oxide responsiveness occurs because of alterations in smooth muscle cell function or structural changes in the artery. During NMD measurement, administration of nitroglycerin, an exogenous source of nitric oxide, allows assessment of nitric oxide responsiveness.
Although previous small studies found reduced FMD in patients with advanced CKD (29–33), there has been less focus on endothelium-independent processes. We found a substantial reduction of NMD in the advanced CKD group compared with the three control groups, and we also found a progressive reduction in NMD with decreasing eGFR.
Three previous small studies with sample sizes ranging from 58 to 106 that compared NMD in patients with CKD and healthy controls did not find statistically significant differences (29,31,33), although there were suggestive trends (33). In a study of vascular function in diabetes, there were only minor differences in NMD between patients with and without diabetes; however, the 26 individuals who were diabetic with overt nephropathy had decreased NMD compared with subjects who were diabetic with normal kidney function (34). In another study of patients who were diabetic, NMD was inversely related to CKD stage and also lower in subjects with microangiopathy (including diabetic nephropathy) than in those without (35). In a study of 11 patients who were normotensive and on hemodialysis, the vasodilatory response to exogenous nitric oxide, assessed using an invasive measurement technique, was decreased compared with 11 healthy controls (36). A recent large multicenter study of patients with advanced CKD found NMD values similar to those observed in this study, but there were no non-CKD controls for comparisons (37). Taken together, the findings from prior work and this study indicate that, in the uremic environment, functional changes in the vasculature are not limited to the endothelium.
CKD-associated pathophysiologic processes that might underlie reduced NMD include arterial calcification (38–40) and oxidant-induced alterations in nitric oxide responsiveness of smooth muscle cells (41). The finding in a small study of patients on hemodialysis that endothelial function but not the vasodilatory response to an exogenous nitric oxide donor improved after kidney transplantation is consistent with fixed structural alterations as the cause of reduced NMD in CKD (42). However, there is also indirect evidence that decreased responsiveness to nitroglycerin (and thus, nitric oxide) in kidney failure is mediated by alterations in either smooth muscle function or smooth muscle cell signaling pathways downstream from nitric oxide (43,44).
Our study has limitations. The cross-sectional nature precludes inferences about causality. The no overt vascular disease group, which was assembled to represent a healthy cohort, may have had some level of vascular disease, because approximately one fifth of the individuals were using an antiplatelet agent. However, this misclassification would tend to underestimate the differences between the healthy and the advanced CKD groups. Additionally, because our exclusion criterion for reduced kidney function in the control groups was on the basis of serum creatinine rather than eGFR, the control groups included some individuals with reduced eGFR, which was evident in our analyses using eGFR quartiles. However, this also would have led to underestimation of the differences between the CKD and non-CKD groups.
The study also has several strengths. The FMD and NMD measurements were performed by a single laboratory using a rigorously standardized protocol. The study was larger than previous studies that included non-CKD controls, and it incorporated strict selection and matching processes for the control groups. Previous studies did not provide information about how control subjects were selected or about their clinical characteristics. The attenuation of differences in FMD between groups after adjustment for demographic and clinical factors illustrates the importance of incorporating careful matching and/or adjustment in these types of comparisons. Our study is unique in its use of three different control groups with graded manifestations of overt vascular disease, enabling us to make observations that are specific to advanced CKD.
In conclusion, the findings of this study suggest that, in advanced CKD, impairment in endothelium-independent vascular function is substantial and greater than that observed in individuals with clinical vascular disease but preserved kidney function. Further work is needed to elucidate the basis for differences in the relative contributions to vascular dysfunction of endothelium-dependent and -independent processes and determine whether targeting endothelium-independent processes, such as smooth muscle function, improves clinical outcomes for individuals with CKD.
Disclosures
None.
Acknowledgments
This work was supported in part by a Research Fellowship from the National Kidney Foundation (T.K.).
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
- Received December 14, 2016.
- Accepted June 9, 2017.
- Copyright © 2017 by the American Society of Nephrology