The Art of Ultrafiltration, from Pump to Peritoneum

Heart failure (HF) continues to present a significant cause of public health concern due to its high prevalence as well as the morbidity, mortality, and financial burden associated with its care [1]. Congestion and in particular fluid overload in patients with HF have been associated with its prognosis, bringing the kidney to the forefront as it maintains a major role in the regulation of fluid homeostasis [2]. Thus, the HF treatment’s efficacy is inextricably linked with its response to decongestive therapeutics [3]. Renal dysfunction may adversely impact potential management options for HF, though an insufficient response to decongestion efforts has been associated with more adverse outcomes in patients with deteriorating renal function [4]. Renal venous congestion has been implicated in affecting the complex interactions of the heart and the kidney [3, 5]. Given sodium’s role as the main factor of extracellular volume, and its role in fluid retention/congestion, sodium extraction rather than that of sodium-free water is an important marker for decongestion [6, 7]. This is further supported by the findings from a secondary analysis of the Renal Optimization Strategies Evaluation-Acute Heart Failure (ROSE-AHF) trial. In patients hospitalized for acute HF, greater urine sodium excretion was strongly associated with reduced 6-month mortality [8]. The so-called sodium non-excretors have been noted in prospective observational studies to have the worst outcomes for 1-year mortality on follow-up [9].

Thus far, it has been loop diuretic therapy that remains the mainstay to treat fluid and sodium retention in acute decompensated HF (ADHF) [10]. This makes a diminished response to the loop diuretic therapy at the forefront detrimentally impacting a patient’s clinical course and prolonging their hospitalization [11]. Therefore, managing the problem of diuretic resistance (DR) is a challenging one to undertake in the inpatient management of ADH. DR is generally taken to mean insufficient response in decongestion despite escalating loop diuretic therapy. A more formal definition for purposes of analyses taking into account urine sodium excretion, DR would be defined as a spot urine sodium content of less than 50–70 mEq/L after 2 h, after and/or an hourly urinary output less than 100–150 mL during the first 6 h after loop diuretic administration in the setting of congestion with volume overload [11]. A review of the widely adopted strategy of combination diuretic therapy (CDT) utilizing the principles of sequential nephron blockade to combat DR has revealed that although there is initial greater decongestion in patients with ADHF receiving CDT that benefit is fairly modest [12‒14].

Assessment of and response to decongestion can be difficult as unfortunately there is not a single accurate measurement to determine volume status. In this realm of clinical assessment of volume status, multiple instrumental modalities have been proposed to supplement physical examination. One already being integrated into undergraduate, graduate, and continuing medical education is the point of care ultrasound defined as the limited bedside ultrasound examination to find answers to focused clinical questions such as volume status [15]. Other than point of care ultrasound, wireless implantable devices permit hemodynamic monitoring systems for select patients and allow physicians to monitor pulmonary artery pressures to guide therapy remotely (e.g., CardioMEMS HF System, Abbot Medical, Abbott Park, Illinois). However, it requires right heart catheterization for implantation and in a large randomized clinical trial did not result in lower composite endpoints of mortality and total HF events compared to the control group [16]. The Blood Volume Analysis (BVA) test, the BVA-100, has also shown promise in establishing quantitative volume methodology based on a recent study showing marked differences in red blood cell mass for chronic HF patients in term of HF-related mortality and hospitalizations [17]. This is particularly important given that recurrent hospitalizations are often associated with worse outcomes and increased costs for HF management given the majority of expenditures for HF is for inpatient care [18].

As it has been established, there remains an unmet need to develop alternative methods of fluid removal with efficacy in patients with HF. The remainder of this review will discuss these potential alternatives broadly defined in the two categories of extracorporeal ultrafiltration (UF) and peritoneal dialysis (PD) therapies.

While HF is typified by the group of cardinal symptoms of dyspnea, fatigue, and fluid retention caused by impaired ventricular filling and ejection failure due to structural or functional heart disease, fluid overload is the key pathophysiological basis for ADHF [19].

The UF process consists of the removal of plasma water from whole blood across a semipermeable membrane or hemofilter in response to a transmembrane hydrostatic pressure gradient generated by a pump [20]. Traditionally the hydrostatic pressure gradient across a hemofilter is created by the classical Starling forces in a capillary bed, in isolated UF however, this hydrostatic pressure gradient is instead generated by an extracorporeal pump in addition to the siphoning effect of suction in the UF compartment [20]. It is a form of extracorporeal renal replacement therapy and is a modification of the conventional hemodialysis procedure which is aimed at isolated fluid removal without providing clearance [21]. The ultrafiltrate consists mostly of water and small and middle molecular weight solutes that are not protein bound so they may pass through the semipermeable membrane into the ultrafiltrate space in response to the transmembrane pressure gradient [20].

During UF therapy, almost all diuretics are typically discontinued to permit the nephrons rest as loop diuretic agents are known to directly increase neurohormonal activation [20]. This diuretic respite helps avoid receptor overexpression, decrease nephron resistance to drugs and thus permitting recovery of diuretic sensitivity [20, 22]. This is with the exception of mineralocorticoid receptor antagonists that may be continued specifically because of their prognostic benefits in HF [22].

The essential parts of a UF circuit or device include venous access to permit the filtration process through the hemofilter as well as the reintroduction of the ultrafiltered plasma into the patient’s systemic circulation. Additionally, a blood pump is necessary to generate the pressure to achieve the UF process either by way of suction applied to the ultrafiltrate compartment (negative pressure) or resistance induced in the venous line (positive pressure). Finally, anticoagulant therapy (e.g., through continuous heparin infusion) to preserve the hemofilter ensures patency of the extracorporeal circuit [23]. This need for anticoagulation is particularly present for continuous UF therapies as well as for the newer more portable UF devices with lower flows to prevent clot formation [24].

Several types of these devices allow for a functioning UF circuit, the first is essentially a modified version of the standard hemodialysis circuits used in the original studies [25]. Such was used in the ULTRAfiltration versus DIureticSon clinical, biohumoral, and hemodynamic variables in patients with deCOmpensated HF (ULTRADISCO) study that utilized PRISMA machines [26]. These traditional devices have the drawback of the need for admission to an intensive care unit (ICU) or cannulation of a central vein and require expert management by nephrologists [20]. Similarly, the initial proof of concept dedicated wearable UF device also was tested on dialysis patients initially given the availability of central venous catheters already in place for dialysis [27]. Since then, there are also newer dedicated UF devices that use peripheral venous access; several landmark studies have been performed using these portable and user-friendly UF machines. Finally, there is the newer miniaturized Artificial Diuresis 1 device that is currently under study and development [28, 29].

Based on the original Italian studies, the mechanistic benefits of UF in HF seem to be rooted in the improved physical performance borne out of the influence of UF on structures around the heart including the pulmonary vasculature, cardiac fossa, and the lungs [30]. By removal of excess fluid from the lungs, UF appears to reverse the observed elevated right atrial and pulmonary pressures along with the increased lung fluid and stiffness thus reducing right ventricular afterload and the work of the heart.

Several clinical trials have helped validate the use of UF in HF over the years. We turn first to the feasibility Relief for Acutely Fluid-Overloaded Patients With Decompensated Congestive Heart Failure (RAPID-CHF) study of 40 patients admitted for CHF with evidence of volume overload. These patients were then randomized to a single, 8-h UF session in addition to pharmacological therapy (PT) or PT alone. It found significant weight loss and fluid removal that was well tolerated in the patients in the UF arm [31]. The investigators concluded that the early application of UF for this patient population was feasible and safe for these patients making way for the following several larger studies on its efficacy.

This brings us to the landmark Ultrafiltration versus Intravenous Diuretics for Patients Hospitalized for Acute Decompensated Congestive Heart Failure (UNLOAD) trial was a prospective, randomized, multicenter trial of UF versus intravenous diuretics for patients admitted with ADHF. At a 2-day follow-up of 200 enrolled patients, the ones randomized to receive UF had greater weight and net fluid loss than the PT group [32]. At 90-day follow-up, the UF group also had fewer patients rehospitalized for HF as well as fewer rehospitalization days for those who had to be admitted to the hospital [32, 33].

That is not to say that every trial in the literature regarding UF in ADHF was not without potential shortcomings. Next, we look at the Cardiorenal Rescue Study in Acute Decompensated Heart Failure (CARRESS-HF) multicenter, prospective, randomized, controlled trial of 188 patients comparing slow continuous venovenous UF to PT and finding that 96 h of UF was associated with a higher rate of adverse events [34, 35]. However, this was related specifically to the perceived effects of UF on kidney function compared to PT. The CARRESS-HF study did not find a significant difference in mortality or weight change of participants on follow-up [35]. The results of the CARRESS-HF trial were potentially confounded by the early termination of the study protocol by substantial dropout and crossover. This hypothesis was supported by the per-protocol analysis of the CARRESS-HF study which was published several years later and found that that UF was indeed associated with higher cumulative fluid loss, net fluid loss, and reduction in weight though yes with higher creatinine and no difference in 60-day mortality [36].

Finally, we turn to the Aquapheresis Versus Intravenous Diuretics and Hospitalization for Heart Failure (AVOID-HF) trial. This trial was designed as a multicenter, 1-to-1 randomized study of 224 patients hospitalized for AHDF who either received adjustable UF or adjustable PT, and found a longer time to first HF event in 90 days and fewer HF and other cardiovascular events for the adjustable UF group [37, 38]. Answering a few practical questions may help further elucidate the role of UF in ADHF and summarize the currently available data on its application in this setting:

• What is the impact of UF on congestion (fluid and weight)?

The findings of the larger studies in this field are collected and tabulated in Table 1. UF provides a greater effect on decongestion as defined by fluid removal and weight loss when compared to standardized PT. Figure 1 depicts the efficacy of CDT and compares it with UF. This increased decongestion in the UF group has been held on subsequent studies including a single-center study of 56 patients [22]. It should be noted that compared to loop diuretics, the fluid removed by UF (i.e., ultrafiltrate) has a much higher concentration of sodium, making UF a more efficient tool for extraction of sodium which is the key determinant of extracellular fluid [39].

• What is the impact of UF on rehospitalization?

The question of rehospitalization rates for those receiving UF compared with standard medical therapy is murkier than others. Some follow-up studies like Continuous Ultrafiltration for Congestive Heart Failure (CUORE) trial of 56 patients admitted with ADHF did find an associated lower incidence of rehospitalizations for HF for UF-treated patients for the following year [40]. The lack of significant difference in reduced readmission rates compared to PT was supported by a recent review of 10 clinical trials comprising a total of 838 patients [45]. However, a Cochrane review examination of this question reports that UF may reduce all-cause and HF-related rehospitalization 30 days from discharge, UF may slightly reduce all-cause rehospitalization and probably reduce HF-related rehospitalizations at the longest available follow-up [25].

• What is UF’s impact on cost?

Due to its more efficient decongestive effect and its potential to lower HF-related rehospitalization days, it is possible for UF to be cost-saving in the long run given potential reductions in length of stay and readmissions despite the higher upfront expenditure. In a recent hospital cost analysis, the investigators reported a 14% cost saving for UF over a 90-day period [46]. The precise impact of UF on HF care cost is yet to be explored at a larger level to determine whether it may indeed offer significant cost savings given multiple confounding factors of whether proprietary cardiology devices or existing nephrological/dialysis infrastructure/support/devices are utilized as well as given how HF readmissions are utilized as a quality metric that can penalize underperforming facilities [1].

• What is UF’s impact on renal function?

Given the findings of the CARRESS-HF study as well as subsequent protocol analysis, there are some questions on the impact on renal function with UF in this patient population. Based on a randomized control trial of 19 patients looking at this question, there was not a significant difference found on 48-h follow-up in patients randomized to receive UF when compared to standard medical therapy [47]. Other studies suggest an initial rise in serum creatinine for patients with intensive volume removal associated with a rise in tubular injury biomarkers; however, these same patients had improved renal recovery and decongestion [48]. In general, the available data support the renal safety of UF in the setting of ADHF. However, since the clinical relevance of the rise in serum creatinine that takes place during decongestion has been debated and the studies of renal injury urinary biomarkers in this setting have been inconclusive, the precise impact of UF on renal function remains to be determined in future studies [49].

• What is UF’s impact on mortality?

On the question of mortality, the data are fairly consistent and represented in the Cochrane review examination of this question that there is the uncertainty of UF compared to PT on all-cause mortality up to 30 days from discharge, and that UF may have little to no impact on mortality on longest available follow-up from discharge [25]. This should be interpreted within the context of the general care of these patients; after successful decongestion and discharge from the hospital, the care of a patient with HF would be affected by a multitude of competing factors that may not be directly related to their volume status.

While PD is traditionally used for the management of patients with end-stage kidney disease, it can also have a role in patients with HF [50]. It is of note that in contrast to hemodialysis where the process of solute clearance can be separated from fluid removal (i.e., providing isolated UF that was discussed above), these two processes are not separable in PD. As such, a patient who receives PD therapy will always have removal of solutes and uremic toxins in addition to fluid extraction; the regimen, however, can be carefully customized to preferentially address one process more than the other one (e.g., predominantly aimed at fluid removal, hence called peritoneal UF or intracorporeal UF). PD for management of HF offers several advantages by avoiding some inherent shortcomings associated with extracorporeal UF. Some of these shortcomings include the cost associated with aquapharetic therapies, worsening anemia due to blood loss and bleeding given the need for continuous anticoagulation, and limited patient access to hemodialysis units [51]. This is where peritoneal UF steps in to offer a choice for daily fluid removal by osmotic UF utilizing different glucose concentrations in the dialysate to customize fluid removal targets [52]. Of course, given the need for a PD catheter placement which is a more involved surgery than the central or peripheral lines necessary for aquapharetic therapy, this mode of UF may be more fitting for those with chronic HF with refractory volume overload rather than patients suffering from an episode of ADHF [53].

The use of PD for refractory volume overload in patients with chronic HF also bears secondary effects of permitting the institution or continuation of some guideline-directed medical therapies for HF. This specifically refers to PD permitting renin-angiotensin-aldosterone inhibition and/or add-on therapies with mineralocorticoid receptor antagonists by modulating the incidence of hyperkalemia [6]. The predominant mechanism for this effect is that PD solutions are free of potassium, so a diffusive gradient helps with the elimination of potassium during PD therapy [6, 54]. This effect is potent and is compounded by the likely insulin release due to the glucose content of PD solutions; patients on PD may even be at risk for hypokalemia and associated adverse health outcomes [55].

We turn now to the HF patient’s perceived quality of life. The importance of validating a HF patient’s reduced quality of life and taking steps to ameliorate it was part of the pilot study implementing a 4-week transition of care program from acute to outpatient setting [56]. This 50-person study found increased weekly education and supportive care including clinic visits with a multidisciplinary team consisting of a nurse practitioner or physician assistant, nurse navigator, pharmacist, social worker, and dietician improved study participants’ quality of life with only 2 patients requiring readmission in 30 days [56]. This is in keeping with the other studies summarizing strategies for improving quality of life often involving input from multidisciplinary outpatient team members that also serve to provide cost savings by reducing readmissions [57]. Further compounding the HF patient’s perceived poor quality of life is the chronic and progressive nature of the disease.

As mentioned above, it is sodium extraction that is a key consideration in driving decongestion in HF, so it is no surprise that sodium removal optimization is to be the focus for PD if it is used for the management of volume in this setting. Water can move from the plasma to the peritoneal cavity by way of small pores and aquaporins at a rate dependent on osmotic force [58]. PD affords customizability of this osmotic force by modification of the prescription of dextrose concentration written for the instilled dialysate though this osmotic force declines with dwell time.

To make use of this, one takes into consideration that each cycle of PD has an early phase where aquaporin activation by crystalloid osmotic gradient results in transcellular removal of sodium-free water, leading to decreased serum sodium concentration of dialysate as the water leaves vascular space to enters the peritoneal cavity by process of sodium sieving [7]. The rapid shift of water alone in the first hour of the instillation of dialysate into the peritoneum via these aquaporins causes a decrease in the concentration of sodium in the dialysate as well as a corresponding increase in the concentration of sodium in the serum. The following phase after this is when sodium transport alongside other small solutes occurs through convection, and diffusion to enter the peritoneal cavity. Sometimes back-diffusion of sodium occurs if dwell times are significantly long through these same small pores of the peritoneal membrane. Volume management is thusly customized by modifying the proportion of these two segments in a dwell either by eliminating the early phase or making the cycles long enough to allow efficient sodium extraction following the early phase [7]. Therefore, if the goal of PD is peritoneal UF for the removal of water and sodium, then the dwell times should be sufficiently short enough in order to maximize water removal via the high transcapillary UF but concurrently long enough to permit sodium the removal of sodium as well. Table 2 summarizes the proposed approaches to enhance sodium removal with PD in patients with HF and this mechanism of peritoneal UF [7].

Multiple clinical trials have been published that support the use of PD in this population of patients with chronic refractory congestive HF. Unfortunately, although there has been an attempt to carry out randomized controlled trials on this topic, it was not completed due to inadequate recruitment [59, 60].

Several observational studies have been conducted including a prospective single center 118 patient study, which found that in patients with refractory chronic HF and chronic kidney disease, PD was associated with improved functional status with regard to NYHA class, significantly improved fluid overload as body weight significantly decreased in this population, and finally, was safe given the relatively low number of PD-related deaths [61].

Another such study was carried out by Courivaud et al. [62] evaluating the impact of PD in a cohort of 126 patients with chronic kidney disease across 2 medical centers in France where those patients with end-stage kidney disease had been excluded. They saw their PD patients experience significant improvement in cardiac function of approximately 10% left ventricular ejection fraction as well as achieving a significant reduction in the number of days of hospitalization for ADHF after PD initiation [62].

A more recent study by Grossekettler et al. [63] that was a prospective, multicenter, and national observational study included data on 159 patients with end-stage HF from a registry. This study also found weight reduction was significantly associated with the use of PD along with significant improvements in NYHA functional class as well as hospitalization rates in these end-stage HF patients treated with PD. The data of these larger observational studies are available to review in tabular form in Table 3.

While much of the discussion thus far on PD in HF has centered around its use in chronic HF, there is some data to support the feasibility of use in some resource-limited settings (including during times as the pandemic where dialysis resources were scarce for otherwise resource-rich areas) for acute HF with concomitant AKI. The 147 patients included in the study mentioned previously presenting as critically ill with acute HF and concomitant AKI received acute PD after placement of PD catheter percutaneously and the subsequent 30 in-hospital mortality were analyzed to be 73.5% [64]. While that is an extremely high mortality rate, it does reveal such an endeavor would be feasible given that PD was able to save some 27% of patients in this study with severe acute HF with stage 3 AKI to at least 30 days. Answering a few questions may help further highlight the role of PD in chronic HF and summarize the currently available data on its application in this setting:

• What is the impact of PD on cardiac function and congestion?

A retrospective analysis of 13 of 21 studies and 537 of 673 patients with refractory HF exploring the impact on left ventricular ejection fraction (LVEF) and New York Heart Association (NYHA) classification reported improvement in LVEF by almost 4.08% with PD as well as improvement in NYHA class from 3.53 to 2.17 on average [51]. Concerning the impact of PD on congestion as measured by sodium extraction, there is also evidence that patients with refractory HF treated with PD have increased sodium removal compared to those who are not [65].

• What is the impact of PD on rehospitalization?

A retrospective analysis of 14 of 21 studies and 416 of 673 patients with refractory HF included in the analysis found that rehospitalization after PD in these patients significantly decreased by 5.08 days per year compared to before PD treatment [51]. Since the majority of admissions to the hospital in patients with HF are driven by congestion and fluid overload, it is not surprising that enhanced sodium extraction by PD is associated with a reduction in the rate of rehospitalization.

• What is the impact of PD on quality of life?

PD offers significant benefits compared to the quality of life for patients in comparison to in-center dialysis modalities as it affords the patient greater control over their therapy as they are empowered to modify their prescription’s osmotic content to titrate the expected fluid removal and given its daily performance, fluid shifts are better tolerated in advanced cardiovascular disease states like HF [66]. Additionally, PD holds an additional quality of life benefit compared to standard PT for HF given the aforementioned benefits in congestion, weight reduction, and rehospitalization rates [60, 67].

• What is the impact of PD on mortality?

While it can be hypothesized that the salutary impact of PD on congestion, sodium removal, and rehospitalization would translate into improved survival, due to a lack of randomized controlled data, it is challenging to draw such a conclusion from available observational studies. However, in a 37-patient study of the use of PD in refractory HF, the use of PD was reported to be associated with reduced mortality rates when compared to standard PT [52].

While much of the data on the use of UF reviewed thus far has been relegated to instances of acute HF and the use of PD for instances of chronic HF, it is important to understand that there are situations where the converse may be true. That is, there may be situations in which it may be more appropriate to use PD in acute HF and UF in chronic HF. There is not much data available on a head-to-head comparison of the use of one modality over the other as an initial intervention as a 2020 systemic review found only four such studies comparing PD to UF with none being randomized controlled studies [68]. The current data for UF look more toward those patients with multiple episodes of hospital admission for acute HF, while for peritoneal UF the data are more for those patients that do not respond well to optimal diuretic therapy in the outpatient with gradual worsening of fluid overload. How these therapies may be contextualized in HF is depicted in the accompanying Graphical Abstract. A proposed a protocol for an UF strategy is further described in Table 4 based upon our personal views on current clinical practices [20].

The interest in the use of PD in the acute setting has seen a rise in recent years with the pandemic coronavirus disease of 2019 (COVID-19) borne primarily out of necessity given the surge of these afflicted patients having concomitant severe AKI and inadequate resources to provide continuous renal replacement therapy or acute dialysis [69]. Prior to the pandemic, the lower infrastructural requirements and lower up-front costs sometimes on the order of 50% of PD compared to HD made it attractive to developing nations [70]. Acute inpatient initiation of PD had been previously explored and due to several factors impacting, its feasibility in clinical practice including the risks of dialysate leaks and peritoneal bacterial infections as well as the relative difficulty in placing a PD catheter compared to HD catheter, it was unable to be recommended as a first-line therapy for acute volume overload or decompensated HF [58]. On reevaluation in the COVID era, PD offered an option for those patients with persistent clotting issues on continuous renal replacement therapy or hemodialysis. These acute patients did pose challenges in being able to obtain access to PD by the traditional route of laparoscopically implanted PD catheter in the operating room, so adaptations needed to be made. Such evolutions during the pandemic included transplant surgeons performing laparoscopic-assisted bedside placement of the flexible PD catheters for ICU or intubated patients or fluoroscopic guided catheter placement by interventional radiology for non-ICU patients [71]. There are some additional considerations to be had in delivering PD to patients in the acute setting who are critically ill as they may also require mechanical ventilation thus necessitating consideration of how changes in intra-abdominal pressure may affect ventilator mechanics. Lowering initial dwell volumes may be indicated in these patients, particularly those with aggressive lung pathologies [69], and limiting the use of PD when patients are prone to avoid ventilator desynchrony [71].

The use of supplemental UF in the populations of patients where PD may be contraindicated or in states of PD failure has been explored as well. Examples of such instances are included in the studies that reviewed the use of acute PD in the COVID-19 era. These include instances when patients previously started on PD are made prone on the ventilator due to severe COVID-19 infection thus limiting PD exchanges to avoid increased intrabdominal pressure, for a positive peritoneal fungal culture, or peritoneal UF refractory volume overload [71].

Fortunately, there are other approaches to the treatment of congestion in cardiorenal syndrome beyond the escalation of loop diuretic therapy and CDT to treat patients’ refractory HF with volume overload. Over the years, the use of UF and PD has shown to be safe and efficacious in facilitating decongestion with pronounced sodium extraction and overall benefits for decreased rates of rehospitalization by leveraging the mechanisms of UF and PD. Future studies including A Randomized Controlled Study to Evaluate the Safety and Effectiveness of the Aquadex System in Patients With Heart Failure and Fluid Overload (REVERSE-HF) currently undergoing recruitment (ClinicalTrials.gov Identifier: NCT05318105). This study looks to be the largest randomized controlled trial of a planned 372 participants looking into a direct comparison of UF versus PT with intravenous loop diuretics on the primary outcome of HF event within 30 days in patients admitted to the hospital with ADHF. The results of this study are expected to further inform our clinical decision-making to provide more efficacious care for this patient population.

Amir Kazory has received consultancy fees from NuWellis, Inc. and Daxor, Inc.

This study was not supported by any sponsor or funder.

B.A: investigation, data curation, writing – original draft, and writing – review and editing. K.A: conceptualization, methodology, validation, supervision, and writing – review and editing. All authors have read and approved the final version.