| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Original Articles |
From the Northwestern University Feinberg School of Medicine, Chicago, Ill (L.K., M.G.); Veterans Affairs Hospital, University of California, San Francisco (B.M.M.); Duke Clinical Research Institute (J.D.L., R.M.C.) and Duke University Medical Center (C.M.O.), Durham, NC; Case Western Reserve University, Cleveland, Ohio (I.L.P.); and University of North Carolina, Chapel Hill (K.F.A.).
Correspondence to Mihai Gheorghiade, MD, Division of Cardiology, Northwestern University Feinberg School of Medicine, Galter 10-240, 201 E Huron St, Chicago, IL 60611. E-mail m-gheorghiade{at}northwestern.edu
Received October 17, 2007; accepted February 4, 2008.
| Abstract |
|---|
|
|
|---|
Methods and Results— In a retrospective analysis of 949 patients from the Outcomes of a Prospective Trial of Intravenous Milrinone for Exacerbations of Chronic Heart Failure, we investigated the relation between admission values and changes in BUN and eGFR and rate of death by 60 days after discharge. On admission, median eGFR was 51 mL · min–1 · 1.73 m–2 (interquartile range, 37 to 70 mL · min–1 · 1.73 m–2), and BUN was 25 mg/dL (interquartile range, 17 to 41 mg/dL). On average, there was a 1.1–mL · min–1 · 1.73 m–2 decrease in eGFR and a 4.7-mg/dL increase in BUN from admission to discharge. By discharge, 12% of patients had a >25% decrease in eGFR, and 39% had a >25% increase in BUN. Although both lower admission eGFR and higher admission BUN were associated with higher risk of death by 60 days after discharge, multivariable-adjusted Cox proportional-hazards analysis showed that BUN was a stronger predictor of death by 60 days than was eGFR (
2, 11.6 and 0.6 for BUN and eGFR, respectively). Independently of admission values, an increase of
10 mg/dL in BUN during hospitalization was associated with worse 60-day survival rate: BUN (per 5-mg/dL increase) had a hazard ratio of 1.08 (95% CI, 1.01 to 1.16). Although milrinone treatment led to a minor improvement in renal function by discharge, the 60-day death and readmission rates were similar between the milrinone and placebo groups.
Conclusions— A substantial number of patients admitted with heart failure have worsening renal function during hospitalization. Higher admission BUN and increasing BUN during hospitalization, independently of admission values, are associated with a worse survival rate. Use of milrinone in these high-risk patients does not improve outcomes despite minor improvements in the renal function.
Key Words: heart failure inotropic agents kidney prognosis renal function risk factors
| Introduction |
|---|
|
|
|---|
Editorial p 2
Clinical Perspective p 33
The blood urea nitrogen (BUN) concentration has been considered a less specific marker of kidney function than eGFR. However, in addition to reflecting GFR, BUN may rise independently of changes in GFR because of enhanced proximal and distal tubular reabsorption under the activation of the sympathetic, renin–angiotensin–aldosterone, and arginine–vasopressin systems.8,9 Although BUN has been associated with adverse outcomes and has previously been incorporated into acute heart failure risk prediction models,10,11 it has not been investigated in patients with decompensated heart failure in conjunction with other measures of kidney function (ie, eGFR). In addition, data on changes in renal function during hospitalization for decompensated heart failure and postdischarge outcomes are scarce.
In a retrospective analysis of the Outcomes of a Prospective Trial of Intravenous Milrinone for Exacerbations of Chronic Heart Failure (OPTIME-CHF), we aimed to determine whether there was an independent increase in postdischarge death and readmission rates observed with higher admission BUN compared with baseline eGFR and with changes in BUN during the index hospitalization for decompensated heart failure.
| Methods |
|---|
|
|
|---|
Renal Function
eGFR was derived from the 4-variable estimating equation from the Modification of Diet in Renal Disease Study: eGFR = 175 x serum creatinine–1.154 x age–0.203 x 1.212 (if black) x 0.742 (if female).14 Using an established definition,15,16 we defined worsening renal function during hospitalization as a
25% decrease in eGFR or a
25% increase in serum BUN from admission to discharge.
Data Analysis
Complete BUN data on admission were available for 936 of the 949 patients enrolled in the study. Continuous variables are presented as median values with interquartile ranges; categorical variables are shown as percentages of nonmissing values. The data are displayed in quartiles. All analyses were performed on the basis of the intention-to-treat principle. Contingent on the assumptions of normality, 1-way ANOVA or the Kruskal-Wallis test was used to explore the relationship between continuous baseline variables and serum BUN or eGFR across quartiles. Pearsons
2 statistics or Fishers exact test was used to examine the association between serum BUN and eGFR quartiles and baseline categorical variables. If there was a significant difference across quartiles, we then examined the significance between the reference quartile and the other 3 quartiles. A similar approach was taken for outcomes except for the primary and secondary end points of the trial. Because of the nonnormality of the primary end point, the Kruskal-Wallis test was used to examine the association between serum BUN and eGFR quartiles and this outcome. For the main dichotomous outcomes measures of in-hospital death and death/rehospitalization to 60 days, logistic regression analysis was used to examine the overall association of outcome and serum BUN or eGFR as continuous variables. If the overall model was significant at the 0.05 level, we examined the significance between the reference quartile and the other 3 quartiles. The relationship between serum BUN or eGFR and death by 60 days was explored by Cox proportional-hazards regression. For this purpose, the follow-up for each individual patient was limited to the first 60 days after enrollment, and the time to death was evaluated within this interval. If the overall model was significant at the 0.05 level, we examined the significance between the reference quartile and the other 3 quartiles. Multivariable analysis was used to assess the independent risk of serum BUN or eGFR on clinical end points. The multivariable analysis model has been described previously11 and included 43 candidate variables that were considered that reflected demographics, cardiac history, comorbidities, bedside clinical assessment, and laboratory studies. Within each category, the most likely predictors were identified through the use of backward-variable selection, and variables were removed from the model at P>0.05. The significant predictors from each of these 5 groups were combined, and 6 prespecified covariates were added if they had not already been included: age, sex, ethnicity, ejection fraction, cause, and systolic blood pressure. These covariates were included on the basis of the high theoretical likelihood that they would be associated with outcomes. Backward stepwise variable selection was performed again, and a model was generated. For this analysis, we then added candidate variables associated with renal function if they were not included in the selective modeling process. This allowed us to adjust for previously known predictors in this study population. When analyzing the relationship between changes in serum BUN and 60-day death rate, we fitted restricted cubic splines to look for linearity in the log-hazards ratio in the Cox model. The effect of milrinone on changes in BUN and eGFR during hospitalization was examined in survivors of initial admission by using linear regression with baseline values as covariates for each measure. Two-sided values of P<0.05 were considered significant for all analyses. Analyses were performed with SAS version 8.2 (SAS Institute Inc, Cary, NC).
The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
| Results |
|---|
|
|
|---|
|
Outcomes by Admission Serum BUN Concentrations
The outcomes of the study patients across serum BUN quartiles are presented in Table 2. The primary end point of days hospitalized for cardiovascular causes within 60 days of randomization was higher in the highest BUN quartile than in the others: 9 days (range, 5 to 24 days), 7 days (range, 4 to 14 days), 6 days (range, 4 to 11 days), and 5 days (range, 4 to 9 days; P<0.01 for each quartile versus the highest), respectively. The in-hospital and raw 60-day death rates also were increased in the highest quartile: 8.9%, 1.3%, 0.9%, and 0% (P<0.01 for in-hospital death) and 22.1%, 6.7%, 4.8%, and 3% (P<0.01 for postdischarge death), respectively. In addition, patients in the highest BUN quartile had higher rates of death or rehospitalizations at 60 days (Table 2).
|
Patients with higher serum BUN concentration also were less likely to reach target dosing of ACE inhibitors by the time of discharge (Table 2).
Baseline quality-of-life scores were comparable across serum BUN quartiles. Although all patients had substantial improvement over baseline in the visual analog of quality of life by hospital discharge that was maintained at the 60-day follow-up, patients in the higher serum BUN quartiles had less improvement at all time points than did patients in the lower BUN quartiles. There was no difference in the baseline quality of life by the subjective health status questionnaire in the 4 quartiles; however, there was a trend for fewer patients in the higher serum BUN quartiles to report feeling better or the same by 60 days compared with patients in the lower BUN quartiles (Table 2).
Outcomes by Admission eGFR
The main clinical outcomes of the study patients across eGFR quartiles are presented in Table 3. The primary end point of days hospitalized for cardiovascular causes within 60 days of randomization was higher in the lowest eGFR quartile than in the other quartiles. The in-hospital and raw 60-day death rates also were increased in the lowest eGFR quartile: 6.8%, 3.4%, 1.3%, and 0.4% (P<0.01 for in-hospital death) and 19%, 10.8%, 3.9%, and 4.3% (P<0.01 for postdischarge death), respectively. Patients in the lowest eGFR quartile also had higher rates of death or rehospitalization at 60 days (Table 3).
|
2, 11.6 versus 0.6 for BUN and eGFR, respectively).
Changes in eGFR and Serum BUN Concentration
In the patients surviving to discharge, there was an average 1.1–mL · min–1 · 1.73 m–2 decrease in eGFR and a 4.8-mg/dL increase in BUN from admission to discharge. Serum BUN increased by discharge in all admission quartiles: 1 mg/dL (interquartile range, –10 to 13 mg/dL), 3 mg/dL (interquartile range, –2 to 11 mg/dL), 3 mg/dL (interquartile range, –2 to 9 mg/dL), and 4 mg/dL (interquartile range, 1 to 10 mg/dL; P<0.001, all quartiles compared with the highest). One hundred six patients (12%) had a >25% decrease in eGFR, and 341 patients (39%) had a >25% increase in serum BUN by discharge. In addition, there were minimal changes in ACE inhibitor doses between admission and discharge, and there was no correlation between changes in ACE inhibitor doses and changes in BUN from admission to discharge (r=0.07).
Multivariable Analysis
The findings of the Cox proportional-hazards model were consistent with those of the unadjusted analysis for the primary end point and for 60-day death rate. After adjustment for baseline differences, serum BUN on admission remained a significant predictor of the number of days hospitalized for cardiovascular causes within 60 days of randomization (P<0.01). Admission serum BUN, when modeled linearly, remained a significant predictor of increased 60-day death rate (hazard ratio [HR], 1.11 per 5-mg/dL increase; 95% CI, 1.07 to 1.15; P<0.01). This translates into an 11% relative increase in the hazard of death within 60 days of hospitalization for every 5-mg/dL increase in admission BUN (Figure 1). Other predictors of death by 60 days were increasing age (HR, 1.26 per 10-year increase; 95% CI, 1.06 to 1.51; P=0.01), lower admission systolic blood pressure (HR, 1.28 per 10-mm Hg decrease; 95% CI, 1.12 to 1.46; P<0.01), New York Heart Association class IV symptoms (HR, 1.93; 95% CI, 1.22 to 3.04; P<0.01 versus New York Heart Association class II to III), and lower admission serum sodium (HR, 1.22 per 3-mEq/dL decrease below 140 mEq/L; 95% CI, 1.04 to 1.42; P=0.01).
|
|
25% worsening in their renal function from admission to discharge between the placebo and milrinone groups. There also were no differences observed in the number of days hospitalized for cardiovascular causes within 60 days of randomization, regardless of the treatment assignment (milrinone or placebo) for any of the baseline serum BUN quartiles. The 60-day death rate also was similar across BUN quartiles in the patients treated with milrinone and those treated with placebo (Table 4). The interaction between treatment assignment and serum BUN was not significant for any outcome (P>0.2), which indicates that the relationship between baseline serum BUN and end points did not differ between treatment groups (Table 4).
|
| Discussion |
|---|
|
|
|---|
In agreement with other studies,1–4,20–23 two thirds of our patients had an abnormal BUN at the time of the admission. The increased in-hospital and postdischarge death rate in our analysis was of a magnitude similar to that described in prior studies.22–24 Interestingly, when we compared the admission BUN and eGFR, BUN was a stronger predictor of death by 60 days after discharge than was eGFR. Similar findings were reported recently in a large Medicare cohort in which a higher BUN on admission to the hospital was a better predictor of postdischarge death than were creatinine-based measures in patients hospitalized for myocardial infarction or heart failure.25
Traditionally, serum BUN concentration has been considered a less specific marker of kidney function than eGFR. However, BUN may rise independently of changes in GFR because of enhanced proximal and distal tubular reabsorption under neurohormonal activation.8,9 Forty percent to 50% of filtered urea is reabsorbed predominantly by the proximal tubules, being linked to the reabsorption of sodium and water under the effects of the renin–angiotensin–aldosterone and β-adrenergic systems.8,9 Increased renal venous pressure in response to increased central venous pressure causes an increase in renal interstitial pressure and in the hydrostatic pressure in Bowmans capsule.26 As result, angiotensin II concentration increases, leading to further sodium, water, and urea reabsorption in the proximal tubule.26,27 In the collecting duct, on binding to V2 receptors, vasopressin increases urea permeability through activation of urea transporters, enabling urea to diffuse into the inner medullary interstitium and increasing its reabsorption and the serum BUN concentration.8,9 In contrast, because creatinine is not reabsorbed but is in fact secreted by the kidney, BUN can rise in the absence of decreases in eGFR.
Although serum creatinine and eGFR are accurate measures of renal function under stable conditions, they may not accurately reflect the totality of underlying processes that occur in the context of decompensated heart failure. In the decompensated state, BUN may be a better prognostic indicator than eGFR, being a marker of a "vasomotor nephropathy" attributed to a transient renal dysfunction related to afferent/efferent arteriolar perfusion mismatch resulting from hemodynamic, neurohormonal, and inflammatory factors.
In outpatients with stable heart failure, worsening renal function occurs at a relatively slow rate of
3% over 6 months.28 In contrast, several studies have shown that 30% to 40% of the hospitalized patients with heart failure will have a >0.3-mg/dL rise in their creatinine level by the time of discharge.4–7,15,21,29,30 In general, this acute worsening occurs within 3 to 4 days of hospitalization and is associated with an increased length of stay and higher risk of in-hospital and postdischarge death.4–7,15,21,29–32 In keeping with the results of the prior studies, 39% of patients in OPTIME-CHF had a >25% increase in BUN by the time of discharge, and 12% had a >25% decrease in eGFR. Although serum BUN may be a better prognostic indicator than creatinine-based measures in decompensated heart failure patients, none of the prior studies have examined the prognostic value of changes in BUN in this population. We found that an increase in BUN during hospitalization, independently of admission values, was associated with worse 60-day survival rate; there was an 8% relative increase in the hazard of death within 60 days of discharge for every 5-mg/dL increase in BUN by the time of discharge.
Although the transient administration of positive inotropic agents is known to improve hemodynamics and has been thought to be one of the most effective ways to mitigate worsening renal function,33 in the present analysis, the use of milrinone was not associated with improved outcomes despite a minor improvement in renal function. However, these findings cannot be extrapolated to other positive inotropic agents without vasodilating and hypotensive effects.
Many prior reports examining the prognostic value of renal dysfunction in hospitalized heart failure patients were small, consisted of retrospective medical records review, or took into account single measures of renal function. The present retrospective analysis is the first to examine this relationship by using data from a prospective, randomized clinical trial that had an extensive standardized assessment of baseline characteristics and outcomes and a 99% complete follow-up rate. In addition, we were able to compare the predictive value of admission BUN and eGFR and to analyze changes in serum BUN between admission and discharge.
The present analysis has several limitations. The study sample is derived from a randomized controlled trial and consists solely of patients with impaired systolic function. Whether serum BUN concentration would be predictive of outcomes in patients hospitalized for heart failure and preserved systolic function deserves further investigation. Patients enrolled in OPTIME-CHF, however, appear to have a clinical profile similar to that of unselected registry patients hospitalized for heart failure who have left ventricular dysfunction.20 Although emerging data suggest that the use of diuretics may be associated with worsening renal function and outcomes in patients hospitalized with heart failure,29,34,35 we did not have complete data on the amount of diuretic use in our patients. Therefore, we cannot comment on the relation between diuretic doses and change in BUN. We chose a specific cutoff to define changes in renal function (ie, 25% worsening) based on prior work in heart failure15 and in contrast-induced nephropathy, the most studied type of renal injury in hospitalized patients with cardiovascular disease16; use of other cutoffs (eg, 0.3-mg rise in creatinine)5 may change the proportion of patients who have worse renal function during a heart failure hospitalization. However, the added prognostic value of changes in BUN should not be affected because we used it as a continuous variable in our modeling. Finally, although higher serum BUN is related to higher risk of death, care must be taken in the interpretation of the exact shape of the relationship because of the relatively small number of deaths observed.
| Conclusions |
|---|
|
|
|---|
| Acknowledgments |
|---|
This investigator-initiated study was supported by Sanofi-Synthelabo, Inc.
Disclosures
Drs Massie, OConnor, Piña, Adams, Califf, and Gheorghiade received grants for the study and/or honoraria from Sanofi-Synthelabo, Inc. In addition, Dr Piña has received research funding from the NIH; has served on the speakers bureaus of A2, GlaxoSmithKline, and Novartis; has received honoraria from A2, GlaxoSmithKline, and Novartis; and has served as a consultant to the US Food and Drug Administration and Sanofi-Synthelabo. Dr Adams has received research funding from and served as a consultant to Novacardia. Dr Gheorghiade has received research funding from the NIH, Otsuka, Sigma Tau, Merck, and Scios; has received honoraria from Abbott, AstraZeneca, GlaxoSmithKline, Medtronic, Otsuka, PDL, Scios, and Sigma Tau; and has served as a consultant to Debbio Pharma, Errekappa, GlaxoSmithKline, PDL, and Medtronic. The other authors report no conflicts.
| References |
|---|
|
|
|---|
10 mg/dL in serum BUN during hospitalization also was associated with worse 60-day survival rate; for every 5-mg/dL increase in the serum BUN from admission to discharge, the 60-day death rate increased by 8% (95% CI, 1.01 to 1.16). Although milrinone treatment led to a minor improvement in renal function by discharge, the 60-day death and readmission rates were similar between the milrinone and placebo groups. These findings underline the significant risk associated with hospitalizations for worsening heart failure and renal dysfunction. This article has been cited by other articles:
![]() |
R. W. Schrier Blood Urea Nitrogen and Serum Creatinine: Not Married in Heart Failure Circ Heart Fail, May 1, 2008; 1(1): 2 - 5. [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Home | Subscriptions | Archives | Feedback | Authors | Help | Circulation Journals Home | AHA Journals Home | Search Copyright © 2008 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |