“Ultrafiltration should not replace diuretics for the initial treatment of acute decompensated heart failure”

Heart Failure (HF) represents a significant and growing health concern in the aging population of the United States. Total HF costs in the US for 2009 are estimated to be $ 37.2 billion, and account for over one million hospital discharges.¹. Acute Decompensated HF (ADHF) represents the most common reason for heart failure hospitalization. Improvements in HF care would thus have a broad impact on health care delivery.

Registry data indicates that the population of patients admitted for HF represents an “at risk” group. Acute in-hospital mortality ranges from 3% and 7% for ADHF and may be as high as 13.5% at 3 months following discharge. Further, surviving patients remain at significant risk for hospital readmission (24%–31%) within three months after their index hospitalization for ADHF. ² Strategies to understand the mechanisms of disease associated with poor outcomes in HF have identified a clinical syndrome of deteriorating renal function, diuretic unresponsiveness, and impaired natriuresis which has been called the cardiorenal syndrome (CRS). Chronic renal insufficiency, commonly associated with HF, adversely impacts heart failure survival, length of stay (LOS) and readmission rates.^{3, 4} While no broadly accepted consensus definition of CRS has been adopted,⁵ most criteria for CRS include: a) HF and renal insufficiency; b) worsening renal function during treatment for ADHF; and c) diuretic resistance.⁶ Worsening renal function (defined by an increased sCr ≥0.3 mg/dl) is a common feature in patients admitted for volume overload and treatment of ADHF, with some reports identifying a prevalence of more than 70% in hospitalized patients. Treatments to mitigate diuretic resistance and CRS have been sought to promote better ADHF outcomes.

An emerging literature suggests an important role for venous congestion as a major contributor to CRS. Traditionally cited mechanisms for worsening renal function include a) systemic and renal hypoperfusion, b) periodic intravascular/arterial volume depletion; c) excessive stimulation of vasoconstrictor neurohormones such as angiotensin, and d) increased interstitial fibrosis associated with the chronic use of furosemide.⁷ However, recent data implicates elevated right-sided venous pressures and increased intraperitoneal pressure due to ascites which commonly accompany right heart failure in the worsening renal function seen in ADHF.^8–10 Thus, acute therapies directed at relieving venous congestion should be paramount in ADHF.

The pharmacological armamentarium for treating symptomatic volume overload has changed very little during the last three decades and remains memorialized in the mnemonic LMNOP frequently learned during medical training to remember acute treatments for heart failure^*. Intravenous (IV) administration of loop diuretics (identified in the “L” of the mnemonic) is effective in decreasing elevated venous filling pressures and remains the consensus first line treatment of volume overload and congestion.¹¹ Nearly 90% of patients in the ADHERE registry receiving IV diuretics for acute congestion during their admission.¹²

Improvement in ventricular filling pressures with diuretics has been shown to be strongly associated with improved survival following hospital discharge.¹³ Loop diuretics are efficacious for the symptomatic relief of volume overload as well as reducing intracardiac and intravascular filling pressures in HF. The usual maneuver to overcome diuretic resistance is to administer increasing doses of diuretics. However, higher doses of furosemide have been linked to higher all-cause mortality rates in retrospective observational studies.¹⁴ More recent data suggest that this association may be incorrect. In a recent prospective analysis of 183 patients with advanced heart failure stratified patients by baseline diuretic dose (furosemide < 80 mg or > 80 mg daily), patients receiving high dose diuretics (n=113) had more markers of increased cardiovascular risk and were more likely to have had a recent history of clinical instablility (33% vs. 4%). After adjusting for clinical stability, diuretic dose was no longer a significant predictor of increased risk.¹⁵ Whether this association is a direct effect of the loop diuretic or simply represents a marker for more advanced heart failure remains uncertain.

UF for acute short-term fluid removal and decongestion in ADHF has been cited in the medical literature for over three decades and represents an alternative to increasing doses of diuretics for decongestion in CRS. However, the FDA approval of the Aquadex system (CHF Solutions, Brooklyn Park, MN) introduced the possibility that UF could be applied routinely in the clinical treatment of ADHF. This system can be used outside the intensive care unit setting, with only peripheral IV access. Advantages of this novel system include: ¹⁶ a) reliable, consistent isotonic fluid removal; b) implementation of a relatively simple prescription/target in patients with known “dry” weight; c) indifference to heart failure mechanism (i.e. systolic vs. diastolic); d) a small amount of extracorporeal blood; and e) implementation using peripheral IV access (as well as others mechanisms reviewed in Table 1A). Several recent clinical studies suggesting possible beneficial effects have stimulated real interest in the HF community for applying this new technology to ADHF management. As the Aquadex system is currently the only approved unit for HF, our discussion focuses on trials using this system. Principal findings of the major published studies using the Aquadex system are summarized in Table 2.

Table 1.

Potential Advantages and disadvantages of Ultrafiltration Therapy: Adapted from Kazory and Ross¹⁶

A. Potential Advantages of Ultrafiltration Therapy
More rapid removal of fluid excess and improvement in symptoms
Higher clearance of sodium load
Isotonic fluid removal
Decreased risk of electrolyte abnormalities (i.e. hypokalemia)
Decreased risk of worsening renal function
Lack of activation of the renin-angiotensin-aldosterone system
Lack of activation of the sympathetic nervous system
Removal of proinflammatory cytokines (with potential restoration of diuretic responsiveness)
Shortened LOS for heart failure hospitalizations
Decreased rate of readmissions for heart failure
More rapid removal of fluid excess and improvement in symptoms

B. Potential Complications/Disadvantages
Significant cost per procedure (device plus disposables)
Additional nursing training and staffing required
Excessive volume removal resulting in hypotension, worsening prerenal azotemia, or acute renal failure
Allergic reaction to the extracorporeal circuit
Catheter-related complications (infection, thrombosis)
Hemorrhage complicating systemic anticoagulation
Hemorrhage from venous return disconnection
Air embolism
Hemolysis and hyperkalemia
Higher clearance of sodium load

Table 2.

Principal Trials and Findings of Ultrafiltration Trials using the Aquadex System. Five trials have been published showing efficacy of UF in heart failure with the Aquadex system. Of these two have been randomized control trials (RCT, boldface) and the remainder have been case series (CS). In these trials, the only finding/outcome supported statistically by more than one clinical trial is fluid removal. E denotes efficacy in achieving a particular goal. When data were not available, these fields were noted as n/a. When a variable was measured but did not differ from control, these data are noted as n.s. Fluid removal, the only parameter that was significantly different in 2 randomized control studies, is highlighted in italics.

Name	Reference	Year	Trial Desgin	Pateints (N)	Symptoms	Weight Loss	Fluid Removal	LOS	Readmission	SCr
SAFE	Jaski (17)	2003	CS	21	n/a	E	E	n/a	n/a	n.s.
RAPID-CHF	Bart (18)	2005	RCT	40	↓*	n.s.	↓*	n.s.	n/a	n.s.
EUPHORIA	Costanzo (19)	2005	CS	20	E	E	E	↓	↓	n.s.
-	Liang (20)	2006	CS	11	E	n/a	E	n/a	n/a	↑*
UNLOAD	Costanzo (21)	2007	RCT	200	n.s.	↓*	↓*	n.s.	↓*	n.s.

^*p <0.05

Jaski and colleagues published the first report¹⁷ of UF accomplished via peripheral IV access with the precursor to the Aquadex system in 2003^†. This clinical series was designed to demonstrate the feasibility of fluid removal with peripheral UF in both inpatients and outpatients. Ultrafiltration was terminated after 1 liter of fluid had been removed or after a total duration of 8 hours of treatment. The primary endpoint of 1 liter of fluid removal was achieved in 23 of 25 treatments (92%). No major adverse events were recorded. Methodological shortcomings included the nonrandomized study design, small sample size, and the lack of data collection/reporting on renal function or clinical outcomes.

The RAPID-CHF trial prospectively tested whether peripheral UF was safe and effective for fluid removal and decongestion in ADHF.¹⁸ This multicenter, controlled trial enrolled HF patients who were randomized to receive either UF or usual HF care including diuretics. UF patients received a single 8 h treatment followed by an initial study assessment at 24 h. Subsequent UF treatments were permitted following the initial assessment. The primary end-point was weight loss at 24 h after enrollment; secondary end-points included fluid removal at 24 h and 48 h after enrollment as well as symptom scores, serum electrolytes and length of hospital stay. Despite a significant difference in fluid removed both at 24 h and 48 h, there was surprisingly no difference between the primary end-point of weight loss between the standard care and UF treatment groups. Subjective dyspnea and CHF scores improved in both treatment groups, although UF-treated patients had a greater improvement in both scores compared to diuretic-treated patients. Additionally, no difference was observed in hospital LOS (6 vs 5 d in UF and standard of care patients, respectively). While this trial did show benefits in volume removed and subjective symptom scores, it failed to meet statistical significance for its primary end point and did not show any difference in hospital LOS.

Costanzo and colleagues published their experience with early and aggressive UF in twenty HF patients in the EUPHORIA study.¹⁹ Inclusion criteria for this uncontrolled observational study included volume overload, a modest degree of chronic renal dysfunction (sCr ≥ 1.5 mg/dl) and a relatively high diuretic requirement at baseline. Key features of this study included: a) the short (<12 h) hospitalization allowed prior to enrollment which resulted in a mean time to initiation of UF of 4.7 ± 3.5 h; and b) the use of continuous UF until ADHF symptoms had resolved, unlike previous studies that allowed only an initial 8 h UF run. In this case series, an average of 8,654 ± 4,205 ml were removed during UF treatment. The average baseline sCr was 2.12 ± 0.60 mg/dl and did not change with UF treatment at discharge or at 90-day follow-up. The mean weight decreased and remained decreased at the 90-day follow-up point. The average duration of hospitalization in this series was 3.7 ± 1.8 days with 60% of patients discharged ≤ 3 days, likely reflecting the aggressive approach to screening and initiating UF. In the 30 days prior to UF treatment, there were 10 admissions for ADHF in 9 study patients; conversely, only one readmission was observed in the study population in the 30 days after UF treatment. Based upon a comparison with historical control from ADHERE, the authors conclude that UF decreases LOS and readmissions. However, the authors' conclusion that readmission rates were decreased after UF by comparing the treatment group to the pre-treatment period, rather than to a randomized control cohort, lacks statistical rigor.

Liang and colleagues presented their experience with 11 serial patients treated with UF at the Mayo Clinic.²⁰ Patients in this report were generally sicker and had been treated in hospital longer than the other UF studies; they had failed at least one IV treatment and were hospitalized an average of 4.4 ± 4.0 days prior to initiation of UF. These authors implemented intermittent UF treatment with the goal of removing 4 liters per 8 hr of UF (a fluid removal rate of 500 ml/h, the maximum of the Aquadex UF system). The 11 patients received a total of 32 UF treatments. The mean baseline sCr was 2.2 ± 0.25 mg/dl and rose to 2.5 ± 0.37 mg/dl following treatment. Strikingly, 45% of this patient population experienced an increase in sCr > 0.3 mg/dl, and 5 of 11 patients received dialysis at the same or during a subsequent hospitalization. The six month mortality rate was 55%, underscoring the baseline severity of illness in the study population. Adverse events associated with the treatment were common and ranged from low flow rates/positional variation in flow (8 of 11 patients) to bleeding complications due to systemic anticoagulation.

The UNLOAD trial, a randomized multicenter controlled trial of 200 ADHF patients that compared UF to standard IV diuretic treatment, represents the single best-designed clinical trial that evaluated UF for the treatment of ADHF.²¹ The primary end points were weight loss and dyspnea score at 48 after enrollment. Secondary end points included net fluid loss and rehospitalization rates. At 48 h, the UF group had lost significantly more weight and fluid than the diuretic cohort. Dyspnea scores did not differ between groups. The HF readmission rate was statistically lower in the UF group (32% versus 18%) as was the number of unscheduled follow-up visits (44% versus 21%). Neither sCr (baseline 1.5 mg/dl) at hospital discharge nor LOS differed between treatment groups. Further, the trial design mandated that the IV diuretic dose be at least double that of the daily outpatient dose during the first 24 hours. Based upon the results reported, the mean diuretic dose was 180 mg/daily while baseline diuretic dose averaged 120 mg daily; these findings suggest a less aggressive approach to fluid removal in this cohort. Finally, information is lacking concerning total rehospitalization rates between the two groups.

The latest American Heart Association/American College of Cardiology practice guidelines fail to recommend UF as a class I therapeutic option for ADHF (see below). This reflects not only the newness of the device but also the absence of substantial randomized controlled data on the safety and efficacy of UF, and argues for a cautious approach to this therapy. The data available to ascertain safety and efficacy are derived principally from only two randomized controlled trial which involved a total of 240 patients with ADHF during a single hospitalization. While the findings from the UNLOAD trial are provocative, difficult questions remain to be answered regarding the appropriate use of UF in the setting of ADHF. An alternative explanation for the apparent benefits observed in the UF group may simply be that the diuretic group received less effective treatment, decongestion, and weight loss. Rehospitalization rates may not have differed had both treatments resulted in a similar degree of volume reduction.

The population of patients admitted with ADHF who are more likely to benefit from UF rather than diuretics is unclear from current clinical trial data. Patients enrolled in most of the UF trials reviewed above gained entry to these studies through a fairly liberal set of entry criteria. In UNLOAD, adult patients within 24 h of admission need only to have exhibited 2 of the following signs of volume overload and congestion: a) peripheral edema ≥ 2+; b) jugular venous distension > 7 cm; c) pulmonary edema or pleural effusion on chest radiograph; d) enlarged liver or ascites; or e) rales, paroxysmal nocturnal dyspnea or orthopnea. Exclusion criteria were more numerous and included among others a) sCr > 3.0 mg/dl, b) SBP < 90 mmHg; c) IV vasopressors; d) vasoactive drug use during or before hospitalization; e) recent use of iodinated contrast; and f) comorbidities expected to “prolong hospitalization”. No specific criteria were developed to select for or identify patients with cardiorenal syndrome. The study population had an average sCr of 1.5 mg/dl (compared with 1.8 mg/dl in the ADHERE registry). ³ Thus, virtually all UF trial subjects were hemodynamically stable with preserved systolic blood pressure and reasonably well preserved renal function at the time of enrollment (the “wet” and “warm” clinical profile). This group (>70–80% of hospitalizations for ADHF) is generally easily decongested with diuretic therapy and has a low in-hospital mortality during conventional treatment. The UNLOAD patients appear to have had better renal function than the usual population admitted with ADHF and it is not yet clear whether the decreased readmission rate can be extrapolated to the more typical and sicker hospitalized HF population.

The population of patients who are most likely to benefit from UF should be better defined. The costs associated with the introduction of any new technology must be carefully evaluated in the current era of cost containment and cost effectiveness analysis. The capital cost for each Aquadex console is approximately $25,000 and disposable supply costs run approximately $900 for a single UF cassette.²² In contrast, the cost of a generic IV diuretic generally averages less than $5/day. The additional costs of UF could be justified if hospital LOS were shortened. However this outcome has yet to be demonstrated in a randomized controlled trial. While the end-point of hospital readmission was decreased in the UF treatment group in UNLOAD, employing UF in the ADHF patient group at greatest risk for readmission would seem to support better the economic case for treatment. However, strategies to prospectively identify this cohort successfully remain in evolution.¹⁶

Beyond decongestion through renal elimination of salt and water, diuretics may exert additional salutary effects including improved cardiovascular performance in acute and chronic heart failure and exert favorable effects of myocardial remodeling. Intravenous furosemide administration results acutely in increased venous capacitance which may be an important factor in acutely ameliorating the symptoms of dyspnea and congestion.²³ Diuretics have also been shown to improve cardiac performance by decreasing afterload. Francis and colleagues found an acute increase then decrease in systemic vascular resistance with bolus IV furosemide administration.²⁴ Wilson and colleagues found an increase in stroke volume and decrease in systemic vascular resistance following more chronic diuretic treatment for ADHF.²⁵ Furthermore, the increased cardiac performance was correlated not with decreased preload resulting from salt and water loss but with lowered systemic vascular resistance. While both diuretics and UF would be expected to improve preload and improve ventricular geometry in functional mitral regurgitation, the additional benefit of decreased afterload may also help reduce the regurgitant flow and improve forward cardiac output. Thus, direct vasoactive effects of diuretics (unrelated to their renal tubular activities) may play a role the benefits obtained with pharmacological therapy.

Chronic administration of diuretics may also favorably affect myocardial remodeling by decreasing myocardial fibrosis. Torsemide (but not furosemide) has been shown to quantitatively reduce myocardial collagen content by endomyocardial biopsy and decrease circulating serum measures of type I collagen synthesis.²⁶ There is also experimental evidence to suggest torsemide can decrease the profibrotic factor, aldosterone.²⁷ However, it remains to be determined whether decreased myocardial fibrosis can result in improvement in diastolic function.

While trials studying the impact of intravenous diuretics on survival in ADHF have not been performed, several studies have been performed using surrogate markers of clinical outcome in heart failure. For example, neurohumoral activation (e.g. plasma endothelin-1, norepinepherine, and B-type natriuretic peptide) decreased rapidly following improvement in ventricular loading conditions produced by intravenous diuretics.²⁸ Unlike loop diuretics, the ability of UF to acutely improve neurohormonal activation remains unproved and the magnitude of the effect (if indeed present) relative to diuretics is unknown. In patients who were acutely congested and in whom an elevated intra-abdominal pressure (IAP) was measured, Mullens and colleagues found a strong correlation between renal function and IAP.⁸ Additionally, the accepted efficacy and safety of intravenous diuretics were reflected in the 2009 update to the ACC/AHA heart failure diagnosis and management guidelines,²⁹ intravenous diuretics represented the only class I recommendation for ADHF. In the setting of inadequate relief of congestion with diuresis, the next recommendation is of intensified diuresis. UF received a class IIa recommendation, recognizing that it is reasonable to apply this therapy in refractory congestion but not as a first line therapy, and that additional studies are needed to define the clinical situations where patients are most likely to benefit.

The potentially deleterious effects of diuretics have been used to argue for a role of UF in ADHF, but efforts are underway to delineate ways to minimize potential toxicities of loop diuretics. Recent data suggest the continuous infusion of loop diuretic results in better diuresis with less likelihood of a decrement in renal function as compared to the bolus diuretic approach that was employed in all published UF trials.³⁰ In fact, this hypothesis is now being tested within the NHLBI-sponsored Heart Failure Network through the DOSE-AHF study. Thus, optimal pharmacologic therapy with diuretics continues to evolve and now includes different administration strategies to produce more effective and potentially less toxic decongestion. A substudy of the UNLOAD trial also argues that both diuretics and UF impact renal physiology similarly. No difference in net fluid removal at 24 hours was observed in this subgroup. Importantly, quantitative measures of renal blood flow, glomerular filtration rate and filtration fraction did not differ between treatments suggesting that equivalent volume reduction results in equivalent effects on renal physiology, and thus defined neither renal benefit nor harm to UF over conventional diuretics, in this small carefully selected population.³¹

The potential for adverse effects and complications of UF therapy should also be considered. In particular, possible adverse renal effects needs to be better evaluated. In the UNLOAD study population, the mean rise in sCr experienced by the UF group at 72 h approached 0.3 mg/dl compared to 0.15 mg/dl in the diuretic group. While the increase in sCr did not achieve statistical significance, the acute deterioration in renal function during volume removal should raise concerns about long-term prognosis. The findings presented by Liang and colleagues at the Mayo Clinic raise the specter that in a high-risk patient population (mean GFR of 38 ml/min which was significantly worse than the UNLOAD population or that reported in either ADHERE or OPTIMIZE-HF) UF may not be the appropriate therapeutic choice for many patients.²⁰ Thus, trials designed to define conditions when UF is safest should also be performed.

Other potential complications associated with UF should also be recognized (summarized in Table 1B). The relatively slow rate of blood flow achieved in the veno-venous circuit can result in filter thrombosis and the need for replacement filters to complete treatment, raising the ultimate cost of UF. To prevent filter thrombosis, systemic anticoagulation with unfractionated heparin or an alternative is required. In addition to increasing bleeding risk, anticoagulation requires additional monitoring of aPTT or ACT. In practice, venous access is often problematic with standard peripheral IV access, requiring the use of a specialized mid-line type catheter or even a central venous line at increased cost and potential risk. Allergic reactions to similar devices have also been described.

Apart from efficacy in volume loss and decreased readmission rate, clinical trial experience with the Aquadex system has not identified endpoints which are consistently improved across published studies (Table 2). Surprisingly, the symptom-derived measurements at the 48 h assessment point in UNLOAD did not differ significantly between the UF and diuretic groups and the weight loss metric was not different between the UF and standard diuretic cohorts in the RAPID-CHF study. The absence of reproducible outcomes across clinical studies is likely multifactorial, representing the relatively small patient populations in each study as well as differences in approach (i.e. intermittent UF at a high rate of volume removal versus a slower continuous rate of withdrawal) to volume removal in each trial. These inconsistencies highlight not only the need to perform additional, larger randomized studies to identify consistent clinical endpoints, but also to study whether the optimal strategy for UF is intermittent or slow continuous removal of fluid. In an attempt to define the safety and efficacy of UF in cardiorenal syndrome, the NHLBI Heart Failure Network and CHF Solutions, Inc. are currently conducting the CARdiorenal REScue Study in Acute Decompensated Heart Failure (CARRESS) study. This multicenter, randomized controlled study aims to enroll 200 people in a 1:1 comparison between peripheral UF and standard diuretic therapy. The primary endpoint is a composite of weight loss and sCr, with numerous pre-specified secondary outcome metrics. Completion of this study will help provide key clinical information regarding the specific application of UF in patients with cardiorenal syndrome.

Finally, as new therapies for heart failure emerge, the end points with which clinical success is measured should be resolved with greater clarity. It is becoming increasingly apparent that treatment strategies which improve symptoms during a hospitalization for ADHF do not necessarily translate in long-term therapeutic success. Tolvaptan, a selective V₂-receptor antagonist, has been shown to decrease symptoms of dyspnea, improve hyponatremia, and promote more effective weight loss than an IV loop diuretic alone during HF hospitalization. However, the recently completed EVEREST trial failed to demonstrate long-term benefits on mortality or other secondary endpoints.³² While the mechanisms of action of UF and vasopressin inhibition differ, the long-term outcomes of any new pharmacologic or device-based therapy must be validated or refuted based upon data from randomized, controlled trials. It is possible the short-term (90 day) reduction in heart failure rehospitalizations observed in UNLOAD may not be sustained or, worse, could be associated with adverse effects on long-term mortality. While lower rehospitalization rates are highly desirable, the standard assessment of any new therapeutic must be its effect on mortality. Thus, a well designed and sufficiently powered outcome trial of UF is needed before this approach should be considered for generalized management of ADHF.

Conclusion

Based upon a critical review of the limited data available in 2009, UF has not consistently demonstrated superiority over aggressive IV diuretic therapy in improving symptoms, weight loss or preservation of renal function and encumbers the potential for additional complications. Proposed benefits such as improved diuretic responsiveness after UF therapy have not been rigorously tested. Significant questions remain regarding the specific populations of patients most likely to benefit from this expensive therapy as well as those in whom it should be avoided. Finally, the optimal strategy (i.e. intermittent vs. continuous) for fluid removal by peripheral UF has yet to be determined. UF represents a promising technology which will likely find a place in the armamentarium of therapies for heart failure. Additional randomized-controlled studies with larger numbers of patients over a broader range of illness are needed. However, until these questions can be answered, UF should not supplant diuretics as first line therapy for routine patients presenting with ADHF.