Potential Effects of Aggressive Decongestion during the Treatment of Decompensated Heart Failure on Renal Function and Survival

Abstract

Background

Overly aggressive diuresis leading to intravascular volume depletion has been proposed as a cause for worsening renal function (WRF) during the treatment of decompensated heart failure. If diuresis occurs at a rate greater than extravascular fluid can refill the intravascular space, intravascular substances such as hemoglobin and plasma proteins increase in concentration. We hypothesized that hemoconcentration would be associated with WRF and possibly provide insight into the relationship between aggressive decongestion and outcomes.

Methods and Results

Subjects in the Evaluation Study of Congestive Heart Failure and Pulmonary Artery Catheterization Effectiveness trial limited data set with a baseline/discharge pair of hematocrit, albumin, or total protein values were included (336 patients). Baseline to discharge increases in these parameters were evaluated and patients with ≥2 in the top tertile were considered to have evidence of hemoconcentration. The group experiencing hemoconcentration received higher doses of loop diuretics, lost more weight/fluid, and had greater reductions in filling pressures (p<0.05 for all). Hemoconcentration was strongly associated with WRF (OR=5.3, p<0.001) whereas change in right atrial pressure (p=0.36) and change in pulmonary capillary wedge pressure (p=0.53) were not. Patients with hemoconcentration had significantly lower 180 day mortality (HR=0.31, p=0.013). This relationship persisted after adjustment for baseline characteristics (HR=0.16, p=0.001).

Conclusion

Hemoconcentration is significantly associated with measures of aggressive fluid removal and deterioration in renal function. Despite this relationship, hemoconcentration is associated with substantially improved survival. These observations raise the question whether aggressive decongestion, even in the setting of WRF, can positively impact survival.

Worsening renal function (WRF) during the treatment of acute decompensated heart failure is a common condition complicating approximately one third of heart failure admissions and is associated with increased length of stay, readmission rate, and decreased short and long term survival (1-4). Historically, mechanisms invoked to explain WRF have centered on hemodynamic perturbations such as reduced cardiac index and overaggressive reduction in filling pressures. However, these theories have not been confirmed in recent publications (5-8). Notably, Mullens and colleagues found that patients with the highest admission cardiac indices and highest discharge right atrial pressure actually had the greatest incidence of WRF (6-8). While these studies have provided strong evidence that diminished cardiac output is not the primary driver of cardio-renal interactions, little can be concluded about the importance of effective intravascular volume depletion as a causative mechanism for WRF. In these publications, right and left sided cardiac filling pressures were the only surrogates for intravascular volume reported. Given that the majority of blood volume is located in the venous system, where pressure and volume have little correlation, the lack of association between WRF and significant reduction in filling pressures does not exclude intravascular volume depletion as a relevant mechanism governing cardio-renal interactions (9).

During diuresis, fluid is directly removed from the intravascular compartment into the urine via the kidneys. The rate at which fluid from the extravascular compartment, such as peripheral edema or ascites, can replace the loss in plasma volume governs the degree of intravascular volume contraction that occurs. If fluid is removed too quickly or to a greater extent than fluid is replaced into the vascular space, than intravascular substances such as red blood cells and plasma proteins concentrate. For quite some time the nephrology community has utilized changes in concentration of blood components, such as red blood cells and plasma proteins, to guide ultrafiltration rates in an effort to avoid intradialytic hypotension and symptoms of intravascular volume depletion (10-12). It has been proposed that these techniques could be extrapolated to the guidance of fluid removal during the treatment of acute decompensated heart failure, potentially allowing the avoidance of WRF and providing a physiologically relevant endpoint for fluid removal (13).

We hypothesized that if intravascular volume depletion is a relevant mechanism in cardio-renal interactions, hemoconcentration would be highly associated with changes in renal function during diuresis. Additionally, we hypothesized that hemoconcentration may more directly represent the aggressiveness of diuresis than surrogates such as changes in filling pressures or deterioration of renal function. Resultantly, hemoconcentration may provide insight into the relationship between the aggressiveness of decongestion and post discharge outcomes, helping to inform the debate over whether WRF is a direct cause of adverse outcomes or simply a reflection of a more advanced disease state.

Methods

The Evaluation Study of Congestive Heart Failure and Pulmonary Artery Catheterization Effectiveness (ESCAPE) Trial was a National Heart, Lung and Blood Institute (NHLBI) sponsored, randomized, multicenter trial of therapy guided by pulmonary artery catheter (PAC) vs. clinical assessment in hospitalized patients with acute decompensated heart failure. Methods and results have been previously published (14,15). Briefly, 433 patients were enrolled at 26 sites from January 2000 to November 2003. Inclusion criteria included an ejection fraction of 30% or less, systolic blood pressure of 125 mmHg or less, and at least 1 sign and 1 symptom of congestion. Exclusion criteria included an admission creatinine level >3.5 mg/dL, use of dopamine or dobutamine >3 ug/kg/min, or any use of milrinone before randomization. Patients were randomized to therapy guided by clinical assessment alone vs. PAC and clinical assessment. Treatment goals were resolution of the signs and symptoms of congestion. In patients randomized to the PAC arm, additional goals of treatment were a pulmonary capillary wedge pressure ≤15 mmHg and a right atrial pressure ≤8 mmHg. Use of inotropes was “explicitly” discouraged. The ESCAPE trial was conducted and supported by the NHLBI in collaboration with the ESCAPE study investigators. This manuscript was prepared using a limited access dataset obtained from the NHLBI and does not necessarily reflect the opinions or views of the ESCAPE investigators or the NHLBI. For the purposes of this analysis, patients with admission and discharge serum creatinine levels (n=401) and at least one baseline/discharge pair of hematocrit (HCT) (n=324), albumin (ALB) (n=157), or total protein (TP) (n=142) laboratory values were included in the analysis. In total 336 patients met these combined criteria. Improved renal function (IRF) was defined as a ≥20% increase and WRF as a ≥20% decrease in estimated glomerular filtration rate (GFR) from baseline to discharge. GFR was estimated using the four variable Modification of Diet in Renal Disease study equation (16). This study was approved by the institutional review committee.

Statistical Methods

Values reported are mean ± standard deviation and percentile unless otherwise noted. Independent Student's t-test or the Mann-Whitney U test was used to compare means of independent continuous variables. Pearson's Chi Square was used to evaluate categorical variables. Paired samples t-test or Wilcoxon Signed Ranks test was used for comparison of continuous variables over time within groups. Between group comparisons of change in admission to discharge values was accomplished by calculating the admission to discharge change in the variable of interest and then applying a Student's t-test or the Mann-Whitney U test. Cox proportional hazard modeling was used to evaluate the univariate hazard ratio of predictors of mortality and rehospitalization. Patients alive or not rehospitalized at the end of follow-up (180 days), respectively, were censored. Variables for multivariable Cox proportional modeling was obtained via entry of all univariate baseline predictors of mortality with a p<0.2 and ≤5% missing values. Using backwards selection, starting with the variable with the largest p-value, variables altering the hazard ratio (HR) by more than 10% were retained in the final model. Proportional hazards models were subjected to 1000 bootstrap replications (with replacement). P values and 95% confidence intervals were derived from these 1000 replications. Additionally, the effect of in-hospital characteristics that could plausibly influence survival was analyzed using forced entry Cox proportional modeling. The proportional hazards assumption was tested by visual inspection of log minus log survival plots. In the case of continuous variables or ambiguous log minus log plots, the proportional hazards assumption was formally tested using time dependent variables representing the product of time and the covariate. The main effects and all covariates were not found to be in violation of the proportional hazards assumption. Statistical analysis was performed with SPSS version 17.0 (SPSS Inc, Chicago, Illinois) and significance defined as 2-tailed p<0.05.

Results

Baseline characteristics of the overall trial population and their interaction with PAC use have been previously reported (15) Characteristics of the 336 patients in the current sub-study are described in Table 1. Additionally, Nohria et al. has previously described the lack of association between PAC derived hemodynamic variables and the development of WRF in this population (8). In the overall ESCAPE trial population there were no significant differences in the rate of WRF, IRF, or mortality between patients that did or did not have pairs of baseline/discharge hematocrit, albumin, or total protein laboratory values available (data not shown).

Table 1.

Patient characteristics

Characteristics	Overall Cohort	No Hemoconcentration (n=102)	Hemoconcentration (n=49)	p
Demographics
Age (years)	56.0 ± 13.6	55.1 ± 13.4	50.7 ± 15.2	0.072
Males	74.10%	79.40%	71.40%	0.276
White race	58.30%	66.70%	53.10%	0.106
Medical History
Ischemic etiology	49.10%	55.90%	36.70%	0.028*
Idiopathic etiology	34.10%	31.40%	32.70%	0.874
Hypertension	45.50%	44.10%	49.00%	0.574
Diabetes	34.40%	35.30%	32.70%	0.749
Atrial fibrillation	31.10%	30.40%	30.60%	0.978
Gout	19.80%	20.60%	22.40%	0.793
Symptoms / Functional Status / Ejection Fraction
Dyspnea at rest	58.80%	52.00%	63.30%	0.191
Fatigue at rest	64.90%	65.70%	7.30%	0.840
Orthopnea	84.80%	83.30%	87.50%	0.509
NYHA class (mean class)	3.9 ± 0.3	3.8 ± 0.4	3.9 ± 0.4	0.604
Six minute walk (feet)	455 ± 422	502 ± 378	641 ± 499	0.156
Maximal oxygen consumption (mL/kg/min)	10.0 ± 3.3	9.5 ± 3.0	9.9 ± 2.8	0.470
Too ill to undergo functional capacity testing	30.50%	15.80%	21.30%	0.419
Ejection fraction (%)	19.3 ± 6.7	19.7 ± 7.4	18.0 ± 5.9	0.150
Admission Physical Exam
Systolic blood pressure (mm Hg)	105.6 ± 16.3	102.9 ± 16.1	105.8 ± 14.7	0.241
Heart rate (bpm)	82.0 ± 15.1	81.3 ± 15.5	86.4 ± 16.7	0.067
Respiration rate (breaths/min)	20.5 ± 4.1	20.0 ± 4.3	19.6 ± 3.7	0.580
Jugular venous pressure > 12 cm	21.50%	24.80%	25.00%	0.974
Rales > 1/3 lung fields	14.70%	4.90%	22.40%	0.001*
Ascites ≥ moderate	19.10%	19.80%	22.40%	0.707
Severe edema	12.00%	10.80%	14.30%	0.534
Medications (baseline)
ACE inhibitor / ARB	89.30%	91.20%	87.80%	0.510
β-Blocker	60.60%	62.70%	59.20%	0.674
Loop diuretic (mg)†	200 (120 to 320)	200 (83 to 320)	280 (160 to 400)	0.005*
Spironolactone	32.00%	36.40%	34.70%	0.842
Thiazide diuretic	12.80%	11.30%	8.50%	0.602
Digoxin	71.70%	74.50%	75.50%	0.894
Medications (in hospital)
ACE inhibitor / ARB	90.80%	92.20%	89.80%	0.628
Loop diuretic (mg)†	240 (120 to 400)	240 (100 to 400)	360 (200 to 480)	0.029*
Thiazide diuretic	30.30%	29.90%	39.60%	0.243
Inotropes	42.80%	51.00%	37.50%	0.124
Vasodilators	29.10%	25.30%	15.60%	0.195
Inotropes or vasodilators	59.30%	62.70%	49.00%	0.108
Laboratory Findings
Hemoglobin (g/dL)	12.5 ± 1.8	12.8 ± 1.6	12.4 ± 1.6	0.218
Hematocrit (%)	37.8 ± 5.2	38.5 ± 4.8	37.4 ± 4.5	0.121
Serum albumin	3.6 ± 0.5	3.8 ± 0.5	3.4 ± 0.5	<0.001*
Serum total protein	7.0 ± 0.8	7.2 ± 0.7	6.6 ± 0.8	<0.001*
Serum sodium (mEq/L)	136.6 ± 4.5	135.8 ± 4.4	137.5 ± 4.0	0.014*
Serum creatinine (mg/dL)	1.5 ± 0.6	1.6 ± 0.7	1.3 ± 0.5	0.011*
Glomerular filtration rate (mL/min)	57.5 ± 25.9	55.3 ± 26.2	67.1 ± 28.1	0.012*
B-type natriuretic peptide (pg/mL)	933 ± 1239	1018 ± 1325	915 ± 1090	0.734
Norepinephrine (pg/mL)	677 ± 537	754 ± 592	766 ± 556	1.000
Treatment Related Parameters
Length of stay (days)	9.0 ± 10.5	8.4 ± 7.7	7.7 ± 3.9	0.378
Change in weight (kg)	3.5 ± 5.0	2.7 ± 3.7	6.3 ± 6.6	<0.001*
Rate of weight loss (kg/day)	0.53 ± 0.86	0.45 ± 0.68	0.95 ± 0.98	<0.001*
Net fluid output (L)	-4.5 ± 5.0	-3.8 ± 4.2	-6.1 ± 6.5	0.039*
Rate of fluid output (L/day)	-0.62 ± 0.67	-0.56 ± 0.64	-0.83 ± 0.74	0.035*

ACE: Angiotensin converting enzyme, ARB: Angiotensin receptor blocker, NYHA: New York Heart Association. Overall cohort represents all patients with any 1 of 3 hemoconcentration variables available in the original ESCAPE dataset. Hemoconcentration was defined as ≥2 of 3 of delta total protein, delta albumin, or delta hematocrit in the highest tertile. P value refers to the difference between those with and those without hemoconcentration.

^*Represents a significant p value.

^†Loop diuretic dose reported as median with interquartile range, all other values represent mean ± standard deviation or %.

In there were no significant changes in baseline to discharge hematocrit (p=0.44), albumin (p=0.84), or total protein (p=0.96). There was no correlation between admission hematocrit (r=-0.05, p=0.34), albumin (r=-0.008, p=0.90), or total protein (r=-0.007, p=0.91) with admission to discharge change in GFR. However, the mean increase in delta ALB (4.1 ± 16.6% vs. -3.4 ± 12.7%, p=0.002) and delta TP (3.1 ± 22.6% vs. -1.9 ± 10.4%, p=0.003) was significantly greater in patients that experienced a reduction in GFR during hospitalization. Mean delta HCT was greater in patients with a reduction in GFR, however, this finding did not meet statistical significance (1.9 ± 12.1% vs. 0.3 ± 12.5%, p=0.055). Changes in blood component concentration were strongly associated with the direction of change in renal function, and the odds for deterioration in renal function were substantially higher for patients in the top tertile of delta ALB, TP, or HCT (Table 2). Receiver operating curve analysis demonstrated delta ALB (AUC=0.813, p<0.001), delta TP (AUC=0.787, p<0.001), and delta HCT (AUC=0.630, p=0.012) to have discriminative ability between patients developing WRF vs. IRF. Notably, the odds for a significant worsening vs. improvement in GFR were 43.5 (p<0.001) times higher in patients with ≥ 2 of the 3 variables in the highest tertile (compared to patients with ≥ 2 of 3 not in the highest tertile). In those with all three measures in the highest tertile (n=22), zero experienced IRF compared to 57.1% in patients with all three in the lowest tertile (n=14, p<0.001). The mean baseline to discharge change in GFR was also significantly associated with hemoconcentration (Figure 1).

Table 2.

Changes in renal function defined by different definitions of hemoconcentration.

	Delta Hematocrit	p	Delta Albumin	p	Delta Total Protein	p	Two of Three	p
Any Deterioration in GFR	1.8 (1.1-2.9)	0.014*	3.4 (1.7-6.8)	0.001*	4.1 (1.9-8.6)	<0.001*	5.3 (2.4-11.7)	<0.001*
WRF	1.4 (0.78-2.4)	0.285	2.7 (1.2-6.1)	0.017*	3.6 (1.6-8.6)	0.002*	5.4 (2.0-14.3)	<0.001*
IRF	0.38 (0.18-0.78)	0.007*	0.05 (0.007-0.39)	<0.001*	0.13 (0.30-0.59)	0.002*	0.06 (0.007-0.44)	<0.001*
WRF vs. IRF	2.9 (1.3-6.8)	0.011*	31.3 (3.7-250)	<0.001*	15.9 (3.2-83)	<0.001*	43.5 (5.2-333)	<0.001*

GFR: Glomerular filtration rate, IRF: Improved renal function, WRF: Worsening renal function. Values represent odds ratio (95% confidence intervals). Delta hematocrit, delta albumin, and delta total protein refer to patients in the top tertile of the respective change in laboratory value. Two of three refers to patients with ≥2 of 3 of delta total protein, delta albumin, or delta hematocrit in the highest tertile.

^*Represents a significant p value.

Figure 1.

Admission to discharge percentage change in glomerular filtration rate grouped by presence or absence of hemoconcentration. GFR: Glomerular filtration rate.

Consistent with previous reports from the ESCAPE trial, PAC derived parameters at baseline, final measurement, or change from baseline to final measurement had no correlation with baseline to discharge change in renal function (p>0.10 for all). Additionally, right atrial pressure at baseline (p=0.53), final measurement (p=0.10), or the associated change (p=0.36) was no different between patients with and without WRF. Similarly, pulmonary capillary wedge pressure at baseline (p=0.32), final measurement (p=0.40), or the associated change (p=0.53) was also not significantly different between groups. The above associations were also not significant for IRF (p>0.5 for all).

Given the strong association between the direction of change in renal function and the presence of ≥2 of 3 measures of hemoconcentration in the highest tertile, these patients were considered to have definitive evidence of hemoconcentration and this definition was used in subsequent analyses. Characteristics of patients with and without hemoconcentration are presented in Table 1. Consistent with a more aggressive diuresis, the patients with hemoconcentration were treated with significantly higher loop diuretic doses, had a greater total weight loss/rate of weight loss, had a larger total volume of diuresis/rate of diuresis, and experienced greater baseline to PAC removal decreases in right atrial pressure and pulmonary capillary wedge pressure (Tables 1 & 3). Measures of disease severity such as hemoglobin, ejection fraction, maximum oxygen uptake, B-type natriuretic peptide level, New York Heart Association class, use of intravenous inotropes/vasodilators, and baseline evidence based medication use was similar between groups (Table 1). Notable exceptions were slightly higher baseline serum sodium and a significantly higher GFR in the group with hemoconcentration (Table 1). However, by the time of discharge these differences were no longer significant (p=0.67 and 0.46 for sodium and GFR respectively). Additionally the group experiencing hemoconcentration had a lower baseline cardiac index, higher baseline loop diuretic usage, and a significantly higher prevalence of baseline pulmonary rales in greater than 1/3 the lung fields. The discharge use of angiotensin converting enzyme inhibitors/receptor blockers (p=0.83), beta blockers (p=0.79), thiazide diuretics (p=0.88), and loop diuretic dosage (p=0.68) was similar between groups.

Table 3.

Pulmonary artery catheter derived variables

Characteristics	No Hemoconcentration (n=54)	Hemoconcentration (n=18)	p
Hemodynamics (baseline)
Right atrial pressure (mm Hg)	11.8 ± 6.7	13.3 ± 7.0	0.397
Pulmonary artery systolic pressure (mm Hg)	56.3 ± 16.2	55.6 ± 9.8	0.808
Pulmonary capillary wedge pressure (mm Hg)	23.8 ± 10.0	27.0 ± 8.4	0.297
Cardiac index (L/min/m2)	2.07 ± 0.49	1.76 ± 0.53	0.038*
Systemic vascular resistance (dyn-s/cm5)	1327 ± 536	1625 ± 562	0.052
Hemodynamics (PAC removal)
Right atrial pressure (mm Hg)	9.3 ± 5.6	6.8 ± 4.2	0.132
Pulmonary artery systolic pressure (mm Hg)	47.1 ± 13.4	42.1 ± 9.4	0.098
Pulmonary capillary wedge pressure (mm Hg)	17.0 ± 7.3	13.6 ± 5.6	0.184
Cardiac index (L/min/m2)	2.38 ± 0.57	2.21 ± 0.39	0.136
Systemic vascular resistance (dyn-s/cm5)	1040 ± 496	1276 ± 364	0.079
Hemodynamics (change)
Right atrial pressure (mm Hg)	-2.6 ± 5.6	-5.4 ± 8.4	0.031*
Pulmonary artery systolic pressure (mm Hg)	-9.0 ± 12.4	-14.4 ± 10.8	0.124
Pulmonary capillary wedge pressure (mm Hg)	-6.2 ± 8.3	-12.6 ± 9.6	0.015*
Cardiac index (L/min/m2)	0.35 ± 0.64	0.46 ± 0.5.3	0.432
Systemic vascular resistance (dyn-s/cm5)	-284 ± 561	-464 ± 563	0.303

PAC: Pulmonary artery catheter. Values represent mean ± standard deviation or %. Hemoconcentration was defined as ≥2 of 3 of delta total protein, delta albumin, or delta hematocrit in the highest quartile. No hemoconcentration was defined as ≥2 of 3 of the above variables not in the top tertile.

^*Represents a significant p value.

Association with Mortality

Consistent with findings in the overall ESCAPE population, WRF did not have a statistically significant association with mortality in the current subset of patients (HR=1.6, p=0.11), (N events=64, 19.3%). Admission serum hematocrit (p=0.39), albumin (p=0.98), and total protein (p=0.28) levels were not associated with survival. Hemoconcentration, however, was associated with a substantially lower risk of mortality at 180 days (HR=0.31, 95% CI 0.055-0.74, p=0.016) (N events=29, 19.3%). After controlling for in-hospital characteristics that could potentially bias mortality estimates (use of inotropes or intravenous vasodilators, in-hospital loop diuretic dose, use of thiazide diuretics, and development of WRF), hemoconcentration remained significantly associated with improved survival (HR=0.25, 95% CI <0.01-0.67, p=0.012) (N events=23, 16.9%). Adjustment for all baseline characteristics associated with mortality with a p≤ 0.2 (initial covariates: GFR, loop diuretic dose, sodium, age, ischemic etiology, idiopathic etiology, significant ascites, severe edema, jugular venous distention greater than 12 cm water, respiratory rate, ACE/ARB use, and fatigue at rest) (final model after elimination: loop diuretic dose, sodium level, significant ascites, severe edema, jugular venous distention greater than 12 cm water, and ACE/ARB use) strengthened the association between mortality and hemoconcentration (HR=0.16, 95% CI 0.02-0.44, p=0.003) (N events=27, 18.6%) (Figure 2). The change in hazard ratio appeared to be largely driven by baseline loop diuretic dose. A model including loop diuretic dose as the only covariate yielded parameter estimates similar to the larger multivariate model (HR=0.15, 95% CI 0.02-0.34, p<0.001) (N events=28, 18.9%). A model including all covariates with the exception of loop diuretic dose produced similar parameter estimates to the univariate hemoconcentration association (HR=0.34, 95% CI 0.05-0.86, p=0.036) (N events=28, 19.0%). Construction of several different models utilizing smaller numbers of covariates also produced results consistent with the importance of loop diuretic dose and the limited impact of non-loop diuretic based covariates on the association between hemoconcentration and mortality. Evaluation of the individual components of the hemoconcentration variable yielded similar overall results (Table 4).

Figure 2.

Survival curves grouped by presence or absence of hemoconcentration after adjustment for baseline characteristics.

Table 4.

Adjusted hazard ratios for death at 180 days for the component variables of hemoconcentration.

	HR (CI)	p
delta HCT	0.58 (0.27-1.03)	0.084
delta ALB	0.33 (0.11-0.97)	0.026*
delta TP	0.14 (0.04-0.50)	<0.001*
Two of Three	0.16 (0.04-0.58)	0.001*

CI: Confidence interval, delta ALB: Baseline to discharge change in albumin, delta HCT: Baseline to discharge change in hematocrit, delta TP: Baseline to discharge change in total protein, HR: Hazard ratio. All HR are after adjustment baseline characteristics (GFR, loop diuretic dose, sodium, age, ischemic etiology, idiopathic etiology, significant ascites, severe edema, jugular venous distention greater than 12cm water, respiratory rate, ACE / ARB use, and fatigue at rest).

^*Represents a significant p value.

Discussion

The principal finding of this study is validation of the concept that the concentration of intravascular substances, such as red blood cells and serum proteins, during the treatment of acute decompensated heart failure provides substantial physiologically relevant information. Hemoconcentration appears to be related to the aggressiveness of diuresis and is associated with higher doses of loop diuretics, greater net weight/fluid loss, higher rate of weight/fluid loss, and greater reductions in right atrial pressure and pulmonary capillary wedge pressure. In patients that experienced hemoconcentration, the odds for deterioration in renal function was substantially increased, lending credence to the theory that intravascular volume depletion is a highly relevant mechanism governing cardio-renal interactions. Additionally, hemoconcentration was associated with a substantially lower risk of mortality, raising the question if aggressive decongestion, even in the setting of worsened renal outcomes, may have a positive impact on survival.

The finding that significant information related to cardio-renal interactions can be obtained from hemoconcentration, when no such data has been found in pulmonary artery catheter derived cardiac filling pressures, is not surprising given that pressure has little correlation with volume in the venous system (9,17,18). The venous system is comprised of the most compliant vessels in the body (approximately 30 times that of the arteries) and >70% of total blood volume resides in the venous compartment (9,19). This great compliance facilitates a relative pressure-volume disconnect allowing large changes in blood volume to be associated with small changes in pressure. In line with this physiology, filling pressures such as central venous pressure and pulmonary capillary wedge pressure have repeatedly demonstrated little to no correlation with measures of volume status such as circulating blood volume and hemodynamic response to fluid challenge (17,19-27). In addition to the inherently compliant nature of these vessels, their capacity can be greatly altered by changes in sympathetic tone with increased tone leading to a significantly altered pressure-volume relationship (9,18,27-29). This is particularly relevant since the final common pathway for changes in renal function during the treatment of decompensated heart failure is thought secondary to neurohormonally mediated decreases in GFR, a form of renal dysfunction often referred to as vasomotor nephropathy (30). Teleologically, it is likely that the same cascade of neurohormonal reflexes activated during volume loss, which serve to defend cardiac output and filling pressures, are also responsible for reduction in GFR and the sodium avid state characteristic of vasomotor nephropathy. Given the remarkable compliance and susceptibility to sympathetically mediated alterations in pressure-volume relationship exhibited in the venous system, it becomes quite understandable how hemoconcentration could be highly associated with changes in renal function while PAC derived filling pressures are not.

Our primary objective in analysis of the association between mortality and hemoconcentration was to discern if the negative association often observed between survival and WRF is secondary to actual harm caused to the patient by aggressive diuresis or if the occurrence of WRF in this setting is simply a reflection of a more advanced disease state. We reasoned that hemoconcentration would likely be tightly coupled to the aggressiveness of decongestion and less so to the disease burden of the patient. This was supported by the very similar overall characteristics of patients with and without hemoconcentration with the exclusion of diuresis-related variables such as loop diuretic dose, weight/volume loss, rate of weight/volume loss, and reduction in filling pressures (all of which were greater in the hemoconcentration group). A strong relationship between hemoconcentration and the direction of change in renal function was observed, however, the polarity of association with mortality was strongly opposed to that associated with WRF. These data suggest that aggressive diuresis is likely not primarily responsible for the excess mortality noted in patients with WRF.

The association between decreased mortality and hemoconcentration is a provocative finding. High dose loop diuretics have been repeatedly associated with increased mortality (31-33). Despite the fact that patients with hemoconcentration were administered higher doses of loop diuretics, they had a profoundly lower mortality. Adjusting for baseline characteristics actually tended to strengthen the association with mortality, suggesting that if anything the group with hemoconcentration may actually have been “sicker.” Taken as a whole, these observations suggest that aggressive decongestion itself may be at least partially driving the association with improved survival and hemoconcentration. It has been proposed that patients discharged with elevated filling pressures have a subsequent greater risk of mortality and congestion may directly participate in the pathophysiologic processes implicated in heart failure progression (13,30,34-36).

Although adjustment for a host of prognostically relevant variables was performed, in an observational study it is impossible to definitively demonstrate that hemoconcentration does not somehow identify an overall healthier group of patients. Significant information related to this question may be gleaned from the ongoing Cardiorenal Rescue Study in Acute Decompensated Heart Failure (CARESS) study and the recently completed Diuretic Optimization Strategies Evaluation (DOSE) trial, both of which compare aggressive strategies of decongestion. If the association between hemoconcentration and improved post discharge mortality are replicated, a randomized trial tailoring therapy based on hemoconcentration may be warranted.

Limitations

The limited number of patients in the ESCAPE dataset with complete sets of laboratory values limits power. While limited power did not affect the primary analysis, likely as a result of the large effect size, associations with smaller effect sizes may not be apparent in this analysis. Additionally, it is possible that some of the modestly significant associations may have arisen as a result of chance due to multiple comparisons and should be interpreted with caution. Moreover, due to the small absolute number of terminal events and the large number of covariates, multivariate models may be unstable and not generalize well to external datasets. The primary purpose for these models was to thoroughly control for baseline characteristics to evaluate the concept that hemoconcentration was not simply a surrogate for healthier patients in this dataset. Additionally, the availability of only baseline and discharge markers of hemoconcentration limits the ability to determine temporal relationships between hemoconcentration and changes in renal function. While the ESCAPE trial is one of the largest contemporary datasets with detailed hemodynamic information regarding decompensated heart failure, by nature of the trial design, PAC data is only available on slightly more than half the patients, limiting power. Additionally, since the treating physicians were not blinded to either renal or PAC data, it is likely that treatment strategies were significantly modified in response to those variables. Moreover, hematocrit, albumin, and total protein are at best imperfect measures of hemoconcentration and superior metrics likely exist. Although strong associations were found between changes in the above markers and several clinical variables/outcomes, the absolute change in these markers was small and thus potentially difficult to apply to individual patients.

Conclusions

Results from this cohort support the notion that reduction in intravascular volume, sufficient to cause hemoconcentration, is strongly associated with changes in renal function during the treatment of acute decompensated heart failure. This concentration of blood components appeared to be related to the aggressiveness of diuresis and suggests intravascular volume depletion may be a highly relevant mechanism governing cardio-renal interactions. Additionally, hemoconcentration was associated with a substantially lower risk of mortality raising the possibility that aggressive decongestion, even in the face of worsened renal outcomes, may impact post discharge survival positively. While these results are internally consistent and in line with plausible pathophysiologic mechanisms, sample size is small and methodologic limitations are significant. As a result these findings should challenge, rather than change, current conceptions and require validation in additional cohorts. Future research is necessary to replicate these findings and identify optimal markers of hemoconcentration. If future investigation replicates these results exploration of the concept of tailored therapy guided by hemoconcentration may be warranted.

Clinical Perspective

Previous studies of acute decompensated heart failure (ADHF) have demonstrated a strong association between worsening renal function (WRF) and increased mortality. Similarly, high dose loop diuretics have been associated with decreased survival and aggressive diuresis has been proposed as a possible cause for WRF. However, it has also been suggested that incomplete relief of congestion during ADHF may contribute to heart failure disease progression and worse survival. The aim of this study was to examine the relationship between aggressive decongestion, WRF, and survival in relation to hemoconcentration. Hemoconcentation occurs when the rate of diuresis exceeds the refill rate of extravascular fluid into the intravascular space, causing concentrations of hemoglobin and plasma proteins to increase. This study demonstrated that hemoconcentration was associated with higher doses of loop diuretics, greater rates and quantities of diuresis and weight loss, greater reductions in cardiac filling pressures, and WRF. However, patients that developed hemoconcentration had substantially improved survival despite higher incidence of WRF. These observations support aggressive decongestion of excess fluid during ADHF, even when high diuretic doses are required and renal function may initially worsen.