Visual Abstract
Abstract
Background and objectives Endothelin-1 is a potent endothelium-derived vasoconstrictor peptide implicated in the pathogenesis of hypertension, congestive heart failure, and inflammation, all of which are critical pathophysiologic features of CKD.
Design, setting, participants, & measurements To test the hypothesis that plasma endothelin-1 levels are associated with increased risks of mortality and hospitalization in patients with chronic kidney failure, we measured plasma endothelin-1 levels in a prospective cohort of 794 individuals receiving maintenance hemodialysis. The primary outcomes were time to death and time to hospitalization.
Results The median plasma endothelin-1 level was 2.02 (interquartile range, 1.57–2.71) pg/ml. During a median follow-up period of 28 (interquartile range, 21–29) months, 253 individuals (32%) died and 643 individuals (81%) were hospitalized at least once. In multivariable models adjusted for demographic, clinical, and laboratory variables, individuals in the highest quartile of plasma endothelin-1 had a 2.44-fold higher risk of death (hazard ratio, 2.44; 95% confidence interval, 1.61 to 3.70) and a 1.54-fold higher risk of hospitalization (hazard ratio, 1.54; 95% confidence interval, 1.19 to 1.99) compared with individuals in the lowest quartile. The Harrell C-statistic of the fully adjusted model increased from 0.73 to 0.74 after addition of natural log-transformed plasma endothelin-1 (P<0.001) for all-cause mortality, and increased from 0.608 to 0.614 after addition of natural log-transformed plasma endothelin-1 (P=0.002) for hospitalization.
Conclusions Higher plasma endothelin-1 is associated with adverse clinical events in patients receiving hemodialysis independent of previously described risk factors.
Podcast This article contains a podcast at https://www.asn-online.org/media/podcast/CJASN/2020_05_15_CJN11130919.mp3
- Endothelin-1
- hemodialysis
- mortality
- cardiovascular disease
- risk factors
- Prospective Studies
- Kidney Failure
- Chronic
- dialysis
- Endothelins
- Renal Insufficiency
- Chronic
- Vasoconstrictor Agents
- hospitalization
- heart failure
- hypertension
- endothelium
- Inflammation
- Cohort Studies
Introduction
Hemodialysis (HD) is a life-sustaining therapy for many patients with advanced stages of kidney disease. Despite progress in HD technologies and medical care, patients on HD remain at markedly increased risk of death, most commonly owing to cardiovascular disease. The reasons for accelerated mortality and high cardiovascular disease risks in the HD population are not fully explained (1–⇓⇓4). Contributing factors include the high prevalence of hypertension, congestive heart failure, inflammation, and volume overload (5).
Endothelin-1 is a potent endothelium-derived vasoconstrictor peptide implicated in the pathogenesis of hypertension, congestive heart failure, and inflammation, all of which are key pathophysiologic features of chronic kidney failure (6). Several epidemiologic studies have demonstrated associations between endothelin-1 and the incidence of congestive heart failure (7,8), pulmonary hypertension (7,9), and CKD (10). Endothelin-1 has also been shown to be independently associated with death and cardiovascular outcomes in disease states such as heart failure (11,12), pulmonary hypertension (9), and coronary artery disease (13). Plasma endothelin-1 levels are increased in patients with CKD compared with healthy individuals, and influence BP regulation (14,15). In patients on HD, smaller studies have shown associations of endothelin-1 levels with ischemic heart disease (15) and other surrogate cardiovascular end points such as intradialytic hypertension (16–⇓18), endothelial dysfunction (19), and incidence of atherosclerosis (20). However, less is known about whether endothelin-1 is associated with clinical outcomes in patients on HD, who have an extraordinarily high prevalence of hypertension and vascular disease. To explore the potential relevance of endothelin-1, we measured plasma endothelin-1 levels in a prospectively assembled cohort study involving patients undergoing prevalent HD. We hypothesized that endothelin-1 levels would associate with a higher risk of mortality and hospitalization.
Materials and Methods
Study Design and Study Population
This was a study of anonymized plasma samples and statistically deidentified clinical data obtained from a biorepository assembled by DaVita Clinical Research (Minneapolis, MN). The DaVita Clinical Research biorepository comprises blood samples and clinical data from 4028 individuals with prevalent chronic kidney failure who received HD at a large dialysis organization between May 2011 and October 2013. A subset of this biorepository was provided by DaVita Clinical Research to S.S.W. at Brigham and Women’s Hospital; the remaining samples and data were distributed to three other academic medical centers for research purposes. The biorepository sampling protocol was reviewed and approved by an institutional review board (Quorum IRB, Seattle, WA) and patients provided written informed consent before the initiation of sample collection. Patients with hemoglobin <8.0 g/dl who were aged <18 years, pregnant, or who had any physical, mental, or medical condition that prohibited the ability to provide informed consent were excluded from participation. The study met minimal risk criteria as set forth and defined at Code of Federal Regulations Title 21 CFR 56.111.
Biorepository Biospecimen and Clinical Data Collection
Under the biorepository study protocol, blood samples were collected from each participant at baseline and, after that, every 3 months for up to 1 year. The majority of samples (57%) were drawn after the long interdialytic interval; this proportion did not vary across quartiles of endothelin-1 (P=0.24). Predialysis blood samples were collected and processed according to a standardized protocol: specimens were shipped on refrigerated packs on the day of collection to a centralized laboratory, where they were aliquoted and stored at −80°C. Specimens with cause for rejection (e.g., unspun tubes, insufficient volume, or thawed specimens) or that were received >48 hours from the time of collection were rejected. Anonymized plasma samples were shipped from the centralized laboratory to the researchers on dry ice at −80°C.
Clinical and HD treatment data for each biorepository individual were collected by the dialysis organization during routine care and were maintained in the organization’s electronic health record. Clinical and HD treatment data were provided to the researchers by DaVita Clinical Research in statistically deidentified form.
Measurement of Endothelin-1
Predialysis levels of endothelin-1 in baseline biorepository plasma samples were measured in duplicates using a commercially available ELISA kit (Quantikine Human ET-1 Immunoassay PDET100; R&D Systems, Minneapolis, MN). We assessed the interassay coefficient of variation using 42 blind-split replicate plasma samples obtained from patients with chronic kidney failure enrolled at Brigham and Women’s Hospital for a separate institutional review board–approved study. The mean interassay coefficient of variation from blind-split replicates was 7%.
Outcomes
The two primary outcomes were time to all-cause mortality and time to the first hospitalization, considered from the time of the first blood draw for each biorepository subject. Information on study outcomes was collected through a patient questionnaire at the day of study enrollment and subsequent linkage to the individual’s medical record. For mortality analyses, patients were censored at the time of kidney transplantation (n=37), transfer to a nonaffiliated dialysis unit (n=31), recovery from dialysis (n=3), and at the end of follow-up (December 31, 2014). For hospitalization analyses, patients were additionally censored from the analysis upon death.
Covariates
All demographic information, clinical conditions, HD prescription data, comorbidities, and laboratory values were derived from the biorepository clinical data set, which incorporated data from the patient’s medical record and the medical/family history survey filled out at the day of study enrollment. All laboratory values stated are pre-dialysis, except that for postdialysis BUN. Ultrafiltration volume was calculated on the basis of the difference between the recorded pre- and postdialysis weights. Baseline covariates were considered as the value in the data set with collection date closest to the date of blood draw for the first sample provided for each biorepository participant; covariates were not time updated. During follow-up, clinical and treatment data for each study participant were collected and reviewed for accuracy and completeness by the treating physician and health care team.
Analytic Approach
All statistical analyses were performed using R version 3.4.2. We expressed continuous variables as means (SD) or medians (interquartile range [IQR], 25th and 75th percentiles). Baseline characteristics were compared across quartiles of endothelin-1, using Kruskal–Wallis test for continuous variables and chi-squared tests for categorical variables. To find the clinical and laboratory factors associated with endothelin-1 levels, we used natural log-transformed plasma endothelin-1 as the dependent variable in univariable linear regression models (with significance set at P<0.05/24=0.002) and a multivariable linear model to identify independent factors. We also explored stepwise selection to find the best predictors for plasma endothelin-1. For stepwise selection, the final model was selected by the smallest Akaike information criterion value. We started with a full model and did not restrict any predictors and the number of predictors during the model selection procedure. Multicollinearity was tested by the variance inflation factor, and homoscedasticity was confirmed by Breusch–Pagan test. We used Cox proportional hazard models to examine the unadjusted and multivariable-adjusted risks of outcomes and modeled endothelin-1 in quartiles (with quartile 1 as the reference group) and as a continuous variable (per 1-unit SD of natural-log transformed plasma endothelin-1). Schoenfeld residuals were used to verify the proportional hazards assumption. The cumulative incidence of death during follow-up was estimated by the Kaplan–Meier method, and the log-rank test was used for univariate analyses. We used Poisson regression models to examine the multivariable-adjusted risks of total hospitalizations. The coefficients were calculated using maximum likelihood estimation and then transformed into incidence rate ratios. Our multivariable adjustment strategy was hierarchical and according to biologic and clinical plausibility, as well as observed correlations of endothelin-1 with factors that could confound the association between endothelin-1 and outcomes. We fitted a series of multivariable models: model 1, adjusted for age, race, sex, body mass index (BMI), systolic BP, diabetes as cause of kidney failure, HD access, vintage, and history of cardiovascular disease; model 2, adjusted for the variables in model 1 plus albumin, ferritin, and single-pool equilibrated Kt/V; and model 3 (exploratory), adjusted for the variables in model 2 plus iron, total iron binding capacity, lactate dehydrogenase (LDH), hemoglobin, white blood cells, platelet, red blood cell distribution width (RDW), parathyroid hormone, phosphate, calcium, and ultrafiltration volume. Effect modification of the association between continuous endothelin-1 and mortality by prespecified covariates (age, sex, race, BMI, BP, hemoglobin, dialysis vintage, ultrafiltration volume, diabetes, and history of cardiovascular disease) was tested by including multiplicative interactions terms in the multivariable model. We tested whether endothelin-1 improved risk prediction by using the likelihood ratio test to compare models with versus without endothelin-1. Complete data sets on variables for outcome analyses were available in 97% of the cohort. We found no appreciable difference in results with multiple imputations, and therefore, missing data were not imputed in the primary analysis. All statistical tests were two-sided, and P values <0.05 were considered statistically significant for time-to-event analyses.
Results
Study Population Characteristics
Data from a total of 794 patients were considered in this study. At baseline, mean age was 60±14 years, median dialysis vintage was 37 months (IQR [25–75th percentile], 18–73 months), and 58% were men. The most common reasons for chronic kidney failure were diabetes mellitus (45%), hypertension (32%), and other glomerular diseases (12%). The median plasma endothelin-1 level was 2.02 (IQR, 1.57–2.71) pg/ml. Participants in higher quartiles of plasma endothelin-1 were younger, had higher systolic and diastolic BP, longer dialysis vintage, and higher ultrafiltration volume (Table 1).
Baseline characteristics of 794 participants in the DaVita Clinical Research biorepository, according to quartiles of plasma endothelin-1
Factors Associated with Endothelin-1 Levels
Cross-sectional analyses of clinical and laboratory factors associated with endothelin-1 levels are shown in Table 2. In unadjusted models, clinical variables that correlated with higher endothelin-1 levels included lower age, hemoglobin, platelet count, and iron saturation; and higher diastolic BP, RDW, serum iron, LDH, and ultrafiltration volume. In the multivariable model, hemoglobin, platelet count, RDW, LDH, and ultrafiltration volume were the strongest independent factors associated with endothelin-1 levels. Results were generally consistent in stepwise selection (Table 2).
Clinical and laboratory factors associated with plasma endothelin-1 concentration
Baseline Endothelin-1 and Mortality
Patients contributed a median of 28 (IQR, 21–29) months of follow-up time to the analysis; 253 individuals (32%) died. The incidence of death was 10.7, 12.3, 16.6, and 24.1 deaths per 100 person-years for the first, second, third, and fourth quartiles of baseline endothelin-1, respectively (Figure 1, Table 3). Higher baseline endothelin-1 levels were associated with a higher risk of death in both unadjusted and adjusted analyses. In the fully adjusted model including clinical characteristics, dialysis adequacy, access, dialysis vintage, ultrafiltration volume, and laboratory values, each 1-unit SD higher in log(endothelin-1) was associated with a 1.46-fold increased risk of death (hazard ratio [HR], 1.46; 95% confidence interval [95% CI], 1.26 to 1.69). The association between endothelin-1 and mortality remained consistent in different clinical subgroup analyses (Figure 2) and was accentuated in patients with BMI<25 kg/m2 (compared with ≥25 kg/m2; P for interaction =0.04), dialysis vintage ≥12 months (compared with <12 months; P for interaction =0.02), and in those with versus without prevalent cardiovascular disease (P for interaction =0.002).
Baseline plasma endothelin-1 concentration was associated with increased risk of death.
Associations of plasma endothelin-1 concentration with death
Baseline plasma endothelin-1 concentration was associated with increased risk of death in different clinical subgroups. For BMI and MAP, the missing value (NA) was 2; for vintage, the missing value (NA) was 16. The fully adjusted hazard ratios are shown as the fold of risk of death for per 1-unit SD increase in natural log-transformed endothelin-1. 95% CI, 95% confidence interval; BMI, body mass index; MAP, mean arterial pressure.
Baseline Endothelin-1 and Hospitalization
Patients contributed a median of 28 (IQR, 21–29) months of follow-up time to the analysis; 643 individuals (81%) were hospitalized at least once. Higher baseline endothelin-1 levels were associated with a higher risk of the time to first hospitalization in both unadjusted and adjusted analyses (Figure 3, Table 4). In the fully adjusted model including clinical characteristics, dialysis adequacy, access, vintage, ultrafiltration volume, and laboratory values, each 1-unit SD higher natural log-transformed plasma endothelin-1 was associated with a 1.15-fold higher risk of hospitalization (HR, 1.15; 95% CI, 1.05 to 1.25). In total, 2642 hospitalizations were observed during the follow-up time. In the fully adjusted model, the incidence rate ratio of initial and recurrent hospitalization was 1.35 (95% CI, 1.19 to 1.54) for the individuals in the highest quartile of plasma endothelin-1 compared with individuals in the lowest quartile (Supplemental Table 1).
Baseline plasma endothelin-1 concentration was associated with increased risk of hospitalization.
Associations of plasma endothelin-1 concentration with time to first hospitalization
Assessment of Discrimination and Model Accuracy
We next compared whether addition of endothelin-1 improved model accuracy for the outcomes of mortality and hospitalization. For all-cause mortality, the Harrell C-statistic of the fully adjusted model increased from 0.73 (SEM 0.02) to 0.74 (SEM 0.02) after addition of natural log-transformed plasma endothelin-1 (P<0.001). For hospitalization, the Harrell C-statistic of the fully adjusted model increased from 0.618 (SEM 0.013) to 0.614 (SEM 0.013) after addition of natural log-transformed plasma endothelin-1 (P=0.002).
Discussion
In this study of individuals on maintenance HD, we found that higher baseline levels of plasma endothelin-1 were associated with both an increased risk of death and an increased risk of hospitalization. The observed associations attenuated but remained significant after adjustment for multiple confounding variables.
Endothelin-1 is a potent vasoactive substance that is primarily released by endothelial and vascular smooth muscle cells under conditions of inflammation, vascular stress, and hypoxia (21–⇓23). Endothelin-1 is implicated in the pathogenesis of hypertension, congestive heart failure, and inflammation, all of which are key pathophysiologic features of chronic kidney failure (6). Endothelin-1 also involves in the physiopathologies of various kinds of CKD, such as lupus nephritis, diabetic nephropathy, and polycystic kidney disease (24). The effects of endothelin-1 are mediated through two receptor subtypes, ETA and ETB. Although sustained vasoconstrictive responses are mediated by receptor ETA, ETB stimulates endothelial-dependent relaxation, clears endothelin-1 from the circulation, and may promote natriuresis (25–⇓27).
Chronic kidney failure is a unique clinical setting notable for exceedingly high rates of morbidity and mortality. Indeed, the risk of cardiovascular disease and mortality in kidney failure patients is 5–30 times higher than in the general population (28). There are several reasons for endothelin-1 to be elevated in patients with chronic kidney failure. Reduced catabolism and clearance by the failing kidneys may contribute, although animal studies have shown that lung and liver metabolism are also important determinants of endothelin-1 clearance (29). The median level in our study was 2.02 pg/ml. In other studies in non-CKD populations, median levels were around 1–2 pg/ml (9,30,31). Upregulation of endothelin-1 has been implicated in distinct pathways in the pathogenesis of the cardiovascular disease, including in hypertension and atherosclerosis, which are common comorbidities among patients with CKD (21,22). Certain clinical hallmarks of kidney failure have been implicated in the increased production of endothelin-1 by endothelial cells. These include endothelial injury and shear stress (32), as well as acidemia (33,34) and venous congestion (35). In animal studies, ETB receptors, which have been implicated in the clearance of circulating endothelin-1 (36), were found to be downregulated in the endothelium in a uremic milieu (26). Increases in endothelin-1 may be maladaptive in kidney failure and contribute to cardiovascular disease pathogenesis and mortality from its vasoactive properties and/or its mitogenic effects on vascular smooth muscle cells. Endothelin-1 contributes to endothelial dysfunction and atherosclerosis, and the higher endothelin-1 may lead to atherosclerosis in patients receiving HD (37,38). Also, arterial stiffness is an independent risk factor for mortality in patients undergoing HD, and endothelin-1 is the major hormone regulating vascular tension in vivo (39). Our finding of statistical interaction with preexisting cardiovascular disease could suggest that endothelin-1 is more maladaptive in those with higher burden of disease and higher risk of mortality. The observed association between endothelin-1 levels and ultrafiltration volume in this study could reflect an increase in endothelin-1 production in volume overload states. This is consistent with prior studies showing a positive correlation between endothelial production of endothelin-1 with mechanical stress (40) and with venous congestion (35). To our knowledge, our findings of associations between endothelin-1 levels and hematologic parameters in patients with kidney failure who are receiving HD have not been reported before, and the reasons for such associations have yet to be fully elucidated. In addition to its potent vasoconstrictive function, endothelin-1 has also been shown to be involved in many inflammatory processes and upregulation of inflammatory cytokines such as TNF-α, IL-1, and IL-6 (41). We observed strong correlations of endothelin-1 with the severity of anemia, lower platelet counts, and lower iron levels, which may reflect the persistent low-grade inflammation in chronic kidney failure, suggesting a potential connection between endothelin-1 and cytokine-induced anemia, impaired iron metabolism, and platelet activation.
Although our findings cannot show causality between endothelin-1 and adverse outcomes in patients on maintenance HD, they raise the question of whether endothelin-1 antagonism could be a promising therapeutic approach in treating patients with chronic kidney failure. Endothelin receptor antagonists have been studied in several clinical conditions, including hypertension, congestive heart failure, pulmonary arterial hypertension, and diabetic kidney disease (42). A common side effect observed in clinical trials of endothelin receptor antagonists has been edema and volume overload from sodium and water retention (43,44). It is worth highlighting that in patients on maintenance HD, volume is primarily regulated through the HD procedure; endothelin-1 antagonists would therefore not be expected to lead to volume retention in most patients undergoing HD, who are typically oligoanuric. Whether this is true would require further study, as endothelin-1 antagonists could affect volume homeostasis through other mechanisms even in patients receiving HD.
There are several important limitations to our study. We attempted to control for common potential confounding variables, but residual confounding may still exist. Patients consented and enrolled in this biorepository are from a single, large dialysis organization (albeit with sites across the United States); as a result, the findings may not be generalizable to the overall HD population. We did not have adjudicated data on the cause for hospitalization or information on dialysis adherence. Thus, we were not able to compare cardiovascular versus noncardiovascular causes nor the influence of treatment adherence on the study outcomes. Whether endothelin-1 levels were higher because of reduced clearance from kidney failure or because of increased production is not clear. Moreover, a causal relationship between endothelin-1 and hard outcomes cannot be established. In our study, endothelin-1 levels were obtained as a single time-point measurement before HD; thus, we could not demonstrate its response to various intradialytic pathophysiologic events. The relationship between endothelin-1 and BP needs further investigation, as prior studies have demonstrated associations between intradialytic hypertension and dynamic changes in endothelin-1 levels that occur during and after HD (17,18,45,46). Furthermore, measurements of inflammatory markers such as C-reactive protein and erythrocyte sedimentation rate were not available in our data set. The relationship between these markers and endothelin-1 requires further investigation.
In conclusion, higher levels of plasma endothelin-1 are independently associated with an increased risk of mortality and hospitalization in patients with kidney failure who are on maintenance HD. Whether endothelin-1 antagonists may be useful treatments for hypertension in patients with kidney failure will require additional studies.
Disclosures
Dr. Opotowsky has received consulting income from Actelion and Novartis. Dr. Waikar has received consulting income from Allena, Cerus, CVS, GSK, Harvard Clinical Research Institute, Janssen, Mass Medical International, Strataca, Takeda, Venbio, and Wolters Kluewer, and has provided expert witness consultation for litigation related to Granuflo, Omniscan, statins, cisplatin nephrotoxicity, and mercury exposure.
Funding
Dr. Waikar is supported by National Institutes of Health grants U01DK085660, R01DK103784, U01DK104308, and UG3DK114915. Dr. Li is supported by Beijing Nova Program grant Z161100004916129.
Acknowledgments
Dr. Waikar and Dr. Li designed the study. Dr. Li and Dr. Sabbisetti performed measurements. Dr. Waikar and Dr. Li analyzed the data. Dr. Li, Dr. Waikar, Dr. Schmidt, Dr. Tio, and Dr. Opotowsky drafted and revised the paper. All authors approved the final version of the manuscript.
Supplemental Material
This article contains the following supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.11130919/-/DCSupplemental.
Supplemental Table 1. Risk of total hospitalizations according to plasma endothelin-1 levels.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
- Received September 6, 2019.
- Accepted March 19, 2020.
- Copyright © 2020 by the American Society of Nephrology
Podcast