First in Human Experience with Peritoneal Direct Sodium Removal using a Zero Sodium Solution: A new candidate therapy for volume overload

Abstract

Introduction:

Loop diuretics have well described toxicities and loss of response to these agents is common. Alternative strategies are needed for the maintenance of euvolemia in heart failure (HF). Non-renal removal of sodium directly across the peritoneal membrane (direct sodium removal, DSR) using a sodium free osmotic solution should result in extraction of large quantities of sodium with limited off target solute removal.

Methods:

This report describes the pre-clinical development and first-in-human proof of concept for DSR. Sodium free 10% dextrose was utilized as the DSR solution. Porcine experiments were conducted to investigate the optimal dwell time, safety, scalability, and to determine the effect of experimental HF. In the human study, participants with end stage renal disease (ESRD) on peritoneal dialysis (PD) underwent randomization and crossover to either a two-hour dwell with one liter of DSR solution or standard PD solution (Dianeal 4.25% dextrose, Baxter). The primary endpoint was completion of the 2-hour dwell without significant discomfort or adverse events, and the secondary endpoint was difference in sodium removal between DSR and standard PD solution.

Results:

Porcine experiments revealed that one liter of DSR solution removed 4.1±0.4 grams of sodium in 2 hours with negligible off target solute removal and overall stable serum electrolytes. Increasing the volume of DSR solution cycled across the peritoneum increased sodium removal and substantially decreased plasma volume (p=0.005). In the setting of experimental HF with elevated right atrial pressure, sodium remoThe animal study was conducted in accordance with insval was ~4 times greater than in healthy animals (p<0.001). In the human proof of concept study, DSR solution was well-tolerated and not associated with significant discomfort or adverse events. Plasma electrolyte concentrations were stable and off target solute removal was negligible. Sodium removal was substantially higher with DSR (4.5±0.4 grams) compared to standard PD solution (1.0±0.3 grams, p<0.0001).

Conclusion:

DSR was well-tolerated in both animals and human subjects and produced substantially greater sodium removal than standard PD solution. Additional research evaluating the use of DSR as a method to prevent and treat hypervolemia in HF is warranted.

Clinical Trials Registration:

URL: https://clinicaltrials.gov Unique identifier: NCT03801226

Introduction:

Congestion is the primary driver of morbidity and hospitalization in most patients with heart failure (HF).^1–4 The symptoms and signs of volume overload largely result from the mechanical consequences of fluid (water) retention. However, it is well known that the primary pathophysiologic driver of water accumulation is actually sodium retention, with fluid passively following the retained salt.⁵ A growing literature has reinforced this physiologic principle finding that sodium removal, rather than water removal, appears to be the most important therapeutic target in HF.^6–9

Loop diuretics represent the mainstay of therapy for congestion in HF but generally result in a dilute urine with highly variable sodium content.^10–12 There are direct toxic effects of loop diuretics such as potassium and magnesium wasting, neurohormonal activation, worsening renal function, adverse renal tubular structural remodeling, and development of resistance to these agents is common. ^13–25 There has been growing interest in non-renal approaches to sodium removal. A somewhat underexplored technology for sodium removal in HF is via the peritoneal membrane. Peritoneal dialysis (PD) is a standard therapy for end stage renal disease (ESRD). However, due to the requirement of standard PD to clear low concentration uremic toxins with variable clearance rates, (i.e., clean the blood) sodium removal is, (by design to prevent excessive removal of sodium) highly inefficient. This inefficiency in sodium removal is primarily driven by the fact that PD solutions contain sodium concentrations that are nearly isotonic to plasma (~132 mmol/L), thus sodium removal is primarily driven by solvent drag (i.e., ultrafiltration) with minimal direct contribution by diffusion.

The purpose of the current line of investigation was to develop a technique for high efficiency direct sodium removal (DSR) across the peritoneal membrane, for potential future use as a non-renal approach to treat and prevent volume overload in HF. The principle hypothesis was that by utilizing a zero-sodium peritoneal solution we could leverage the large concentration gradient between extracellular fluid and the solution to drive sodium removal by not only ultrafiltration, but also directly via diffusion. Here we present the pre-clinical development in addition to the results of a first-in-human proof of concept study for DSR.

Methods:

The data, analytic methods, and study materials will be made available upon request to other researchers for purposes of reproducing the results or replicating the procedure. Please address requests for data to the corresponding author.

Several phases of animal experiments were conducted to develop the DSR approach and are described herein. 1) Exploration of kinetics: These experiments explored the overall kinetics of ultrafiltration and sodium removal in healthy swine (n=5) to choose an optimal dwell time. 2) Data generation: Once an optimal dwell time was determined, healthy swine (n=10) were evaluated for effectiveness and consistency of the DSR approach. 3) Scalability: Four of the data generation animals underwent cycling with large volumes of DSR to determine if sodium removal would scale with larger peritoneal DSR exposure. 4) HF model: A right sided congestion animal HF model was developed and the effects of elevated right sided filling pressures on sodium removal was determined in n=5 swine. We next turned to the single dose DSR first-in-human proof of concept based on the above preclinical experiments.

Animals and surgical preparation:

The animal study was conducted in accordance with institutional guidelines and the protocol was approved by the Yale University Institutional Animal Care & Use Committee. Studies were conducted in male farm pigs (~80kg) in accordance with the Animal Welfare Act. Animals were induced with a combination of propofol, diazepam, ketamine, and xylazine, then underwent endotracheal intubation with mechanical ventilation with maintenance anesthesia using isoflurane +/− nitrous oxide. Venous and arterial access were obtained by femoral and jugular cutdowns. To maximize our ability to drain the peritoneum quickly and completely, four standard PD catheters (Medtronci, Inc; Minneapolis, MN, USA.), one in the most dependent position of each of the 4 quadrants of the supine animal, were inserted. These catheters were placed percutaneously with the assistance of a mini-laparotomy, which was closed in layers. In the experimental HF animals, a balloon tipped Swan-Ganz catheter was placed into the pulmonary artery and a fenestrated catheter was placed in the pericardium under direct visualization through a left lateral thoracotomy.

Exploration of kinetics:

After surgical instrumentation, 2L of 6% hydroxyethyl starch in 0.9% sodium chloride (NaCl) was administered as a bolus to prevent volume depletion, prior to beginning the experimental protocol. Next, sodium free 10% dextrose was instilled into the peritoneum. 2.5–25 μCi of I-131 albumin (Daxor Inc; New York, NY, USA) was added to the 10% dextrose to allow for calculation of serial intra-peritoneal volumes and solute content. 5 ccs of peritoneal fluid was sampled every 15 minutes for the first 90 minutes, every 30 minutes for the next 90 minutes, then hourly thereafter. At 300 minutes, fluid was drained. No formal a priori rules were set to choose the dwell time for subsequent experiments, rather choice of time was a subjective decision aimed to optimize future animal experiments, first in human study, and ultimate future patient burden and efficacy. Similarly, no experiments were done to determine the optimal volume but rather 1L was chosen empirically to balance volume reduction from standard PD volumes and ensuring adequate surface area contact to the peritoneum.

Data generation:

After determining the optimal ultrafiltration (UF) and sodium removal kinetics in the above experiments, n=10 swine were instrumented as described above and underwent a 2-hour dwell of sodium free 10% dextrose. Fluid was removed at 2 hours with the assistance of vacuum with determination of total volume and total sodium content.

Scalability experiments:

In 4 of the animals from the data generation experiments (after completing the 2-hour dwell with 1L of 10% dextrose), an additional 10L of sodium free 10% dextrose was cycled in 2.5L increments with 1.5-hour dwell times, similar to the cycling approach in human PD patients. Cumulative sodium removal was determined. Plasma volume was determined by indicator dilution using I-131 albumin (Daxor Inc; New York, NY, USA.) at baseline and again at completion of the procedure. Total sodium removal was calculated from the cumulative drained volume resulting from the four 2.5L cycles.

Experiments with elevated right sided filling pressures:

The goal of these experiments was to understand if elevated right sided pressures (and thus peritoneal capillary pressures) would alter the sodium removal of the DSR procedure. On a population level, acute decompensated HF is primarily a disease of congestion rather than low cardiac output, thus our objective was to select a model where we could achieve elevated right sided filling pressures without severe reductions in blood pressure or cardiac output.^1–4 A cardiac tamponade model was selected and optimized to allow for titratable increases in right atrial pressure without severe reduction in cardiac output or blood pressure. Induction of tamponade was initiated with intravenous administration of 2L of 6% hydroxyethyl starch in 0.9% NaCl and 2L of lactated Ringer’s solution to establish intravascular volume overload (1:1 Normal saline and lactated Ringer’s results in an approximately isonatraemic solution to swine plasma). Next, the pericardium was pressurized to 20–22.5 mmHg using 6% hydroxyethyl starch (to prevent reabsorption of the fluid) in 0.9% NaCl diluted with 10% v/v iohexol iodinated contrast (allowing visualization on fluoroscopy, supplementary figure 1). A pericardial pressure of 20–22.5mmHg was selected as this resulted in blood pressure and cardiac output relatively similar to the pig’s baseline values, but right and left sided filling pressures consistently above 20 mmHg. Right atrial pressure was maintained at 20–22.5 mmHg throughout the experiment using intravenous 6% hydroxyethyl starch in 0.9% NaCl and lactated Ringer’s in 1:1 ratio. After establishment and stabilization of the tamponade, a 2-hour dwell identical to the data generation experiments was then performed.

Single dose first in human study:

Patients receiving PD for ESRD with functioning PD catheters underwent randomization and crossover to open label DSR solution (sodium free 10% dextrose) or standard PD solution (Dianeal Low-Calcium with 4.25% dextrose, Baxter; Deerfield, IL, USA), each separated by 1 week. This Phase 1 study was conducted in prevalent PD patients rather than normal subjects due to the risk that PD catheter placement would pose to a normal subject. Inclusion criteria were 1) Patients actively undergoing PD with a functioning PD catheter; 2) PD vintage < 3 years; 3) Age >18 years of age; 4) As judged by treating nephrologist to be at or above optimal volume status (i.e., not dehydrated). Exclusion criteria were 1) Uncontrolled diabetes with frequent episodes of severe hyperglycemia; 2) Systolic blood pressure <100 mmHg; 3) Serum sodium < 130 mEq/L; 4) 1 or more episodes of peritonitis in the previous 6 months or active infection of the PD catheter; 5) Anemia with hemoglobin <8g/dL; 6) Serum bicarbonate < 18 mEq/L; 7) Anuric renal failure; 8) Inability to give written informed consent or follow study protocol; 8) Pregnant or lactating.

4.25% dextrose PD solution was selected as the comparison solution given it is the most effective marketed PD solution for fluid/sodium removal and has a similar osmolarity to 10% dextrose. Prior to instillation of the study fluid, there was a 30-minute drain of the abdomen with the patient assuming multiple positions during this time to ensure as complete drainage as possible. Given this was a dialysis population with a tendency to develop acidosis, all patients were given 30mEq of sodium citrate/citric acid by mouth. Next, one liter of either DSR solution or standard PD solution was infused into the peritoneum and left to dwell for 2 hours. The intraperitoneal volume was determined longitudinally using indicator dilution technique with I-131 radiolabeled albumin (Daxor Inc; NY, New York, USA), in addition to direct measurement of drained fluid at the end of the dwell. Vital signs, blood (every 30 minutes), and peritoneal fluid (every 15 minutes) were obtained serially throughout the protocol. Patients in the DSR group were given 50% of the UF volume back at the end of the dwell in the form of intravenous normal saline to replace sodium/volume losses. The primary endpoint was safety/tolerability defined as completion of the 2-hour dwell without significant discomfort or adverse event. The secondary efficacy endpoint was the difference in sodium removal between DSR solution and standard PD solution. This protocol and informed consent form were approved by the Yale University Institutional Review Board and all patients gave written informed consent. This trial was registered with ClinicalTrials.gov (NCT03801226) and with the FDA (IND141103).

Assays and calculations:

An Imola (Randox, Crumlin, County Antrim, Northern Ireland) fully automated chemistry autoanalyzer was used to determine peritoneal fluid electrolyte and chemistry values. Blood levels of solute and electrolytes were determined using an iSTAT device (Abbott; Princeton, NJ, USA). I-131 concentrations were determined by gamma counting in either a Cobra Gamma Counter (Canberra-Packard Corporation; Schwadorf, Lower Austria, Austria) or a Daxor BVA device (Daxor Inc; New York, NY, USA). In the porcine experiments, peritoneal volumes were calculated by duplicate counting of a 1 ml sample of I-131 spiked 10% dextrose. Using this concentration and the known volume of 1L of administered solution, subsequent intra-peritoneal volumes were calculated based on the reduction in counts. Since the human subjects only had one PD catheter and an unknown volume of residual peritoneal fluid pre-instillation of study solution was expected, the baseline time zero concentration of I-131 was determined by a linear time zero regression using the first 3 data points, with subsequent volumes calculated using this time zero concentration and known volume of infused solution. The absolute quantity of intra-peritoneal solute was calculated using the I-131 albumin derived volume multiplied by the solute concentration in the peritoneal fluid. Blood volume, plasma volume, and red cell mass were determined on the BVA 100 semi-automated blood volume machine (Daxor Inc; New York, NY, USA) by I-131 albumin indicator dilution and spun hematocrit values. Serum sodium values were corrected for glucose by subtracting 2 mEq/dL for each 100 mg/dL increase above 100 mg/dL.^{26, 27} All reported ultrafiltration volumes are the net ultrafiltration which is calculated as: [UF volume= total drained peritoneal volume - volume instilled].

Statistical Analysis

Descriptive analysis and statistical tests were performed using SPSS, version 24 (IBM; Armonk, NY, USA), SAS software, version 9.4 (SAS Institute Inc; Cary, NC, USA), and Stata version 13.1 (Statacorp; College Station, Texas). Data with a normal distribution are presented as mean ± standard deviations. Categorical values are presented as frequencies and percentages and data with a skewed distribution are shown as median with interquartile ranges (IQR). For animal experiments, independent t tests were used for comparison. For crossover trials with patients, we examined the difference on the biological outcomes between two interventions over time using linear mixed models accounting for correlations within-subjects. The interactions between time (categorized into baseline, 30, 60, 90, and 120 mins) and intervention (DSR vs. PD) were tested for the intervention effects at a 5% significance level. To account for crossover design effect, the linear mixed models were adjusted for the interaction between time and the order of the interventions (coded 1 for DSR following standard PD solution and 2 for PD solution following DSR). No significant crossover effects were found.

Results:

Exploration of kinetics:

Five animals underwent a 6-hour dwell of 1L of 10% dextrose. Contrary to the expectation of early peak UF with fluid reabsorption by 6 hours, we found ultrafiltration continued throughout the entire dwell (Figure 1). Although the glucose osmolarity in the solution became hypotonic to plasma in under 2 hours, the rapid entry of non-glucose osmoles replaced the diluted/absorbed glucose osmoles resulting in the DSR solution remaining hypertonic to plasma for the entire 6-hour dwell (Figure 1). The total ultrafiltration volume was 1.4 ± 0.3 L and the total sodium removed was 5.6 ± 0.2 g. More than half of the total sodium removal occurred in the first 2 hours of the experiment, thus a 2-hour dwell was selected as the dwell time going forward.

Figure 1:

Ultrafiltration volume, sodium content of peritoneal fluid, and peritoneal fluid osmolarity in 5 pigs during a 6-hour dwell with sodium-free 10% dextrose.

Data generation animals:

Ten animals underwent a 2-hour dwell with sodium free 10% dextrose. Total UF volume was 0.91 ± 0.15 L and total sodium removed was 4.1 ± 0.4g. The total glucose absorption over the 2-hour dwell was 51 ± 5% of the administered glucose. Plasma glucose levels increased modestly, and serum sodium levels were overall stable (Supplementary Figure 2). Non-target solute removal was negligible with 5.5 ± 1.0 mmol of potassium, 1.8 ± 0.3 mmol of calcium, 0.8 ± 0.1 mmol of magnesium.

Scalability and ability to reduce intravascular volume:

Four of the above animals underwent cycling with large volumes of DSR solution to determine if sodium removal would scale with larger DSR administration and thus impact intravascular volume. Three of the 4 animals were able to tolerate the full 10L of DSR cycling, one had severe hypotension and was euthanized after the third cycle. In total, an average of 19.4 ± 3.5 grams of sodium was removed. This fluid and sodium removal resulted in substantial reduction in total blood volume and plasma volume (Figure 2).

Figure 2:

Total blood volume, plasma volume, and red blood cell volume in n=4 pigs before and after cycling 10L of sodium free 10% dextrose across the peritoneal membrane.

Post cycling the hematocrit had increased from 23.8 ± 1.0% to 54.3 ± 7.2%. Degrees of freedom = 3 for all comparisons and t value = 7.41 for blood volume, 10.27 for plasma volume and 0.43 for RBC volume. RBC: red blood cell.

Effect of increased right sided pressures on sodium removal:

Fluid loading and pressurization of the pericardium were successful in creating an acute model with relatively preserved cardiac output and blood pressure (Figure 3A) but elevated filling pressures with right atrial pressure and pulmonary capillary wedge pressure consistently above 20mmHg for the duration of the experiment (Figure 3B). Animals underwent the same 2-hour dwell protocol as the normal animals with 1L of sodium free 10% dextrose. Both the degree of ultrafiltration and sodium removal were dramatically increased by approximately 4-fold in the setting of acute right sided filling pressures (Figure 4)

Figure 3:

Hemodynamic parameters of the cardiac tamponade acute heart failure model

After fluid loading (marked “post-fluid”), pigs underwent pressurization of the pericardium (marked “post tamponade”). CO: cardiac output, SBP: systolic blood pressure, MAP: mean arterial pressure, PCWP: pulmonary capillary wedge pressure, CVP: central venous pressure, Pericardial: pericardial pressure.

Figure 4:

Ultrafiltration volume and sodium removal during a 2-hour dwell in healthy animals and animals with experimental heart failure with elevated right sided filling pressures.

Degrees of freedom=4.2 t =9.25. HF: Heart failure.

Human single dose proof of concept study:

Ten patients completed the crossover study (Supplementary Figure 3) and their baseline characteristics are presented in Table 1. The primary endpoint, defined as completion of the 2-hour dwell without significant discomfort or adverse event, was met in all 10 patients. Overall the treatment was well tolerated with 2 of 10 patients reporting mild and short duration cramping during instillation of the 10% dextrose solution, one of which also had similar cramping during the standard PD solution instillation. There were no significant differences in blood pressure or peak heart rate between the two groups (Supplementary Figure 4). Changes in plasma glucose were larger with the 10% dextrose solution compared to standard PD solution with the most pronounced differences early in the dwell (Figure 5). However, differences in plasma glucose completely resolved after draining the solution (Figure 5). There were no patients that developed severe hyperglycemia at any time point in either group (Figure 5). The relative glucose absorption was the same between 10% dextrose and standard PD solution, however given the larger absolute amount of glucose in 10% dextrose the absolute quantity of glucose absorbed was larger with 10% dextrose (Supplementary Figure 5). Serum sodium was not different between groups (Figure 5). Removal of off-target non-sodium electrolytes with DSR such as potassium (5.5 ± 1.1 mmol), magnesium (1.7 ± 2.5 mmol), phosphorus (1.9 ± 0.6 mmol) and calcium (1.6 ± 0.3 mmol) was negligible, and plasma electrolyte and chemistry parameters were stable throughout the dwell (Supplementary Figure 4).

Table 1:

Baseline characteristics

Characteristics	All patients
Demographics
Age, years	54 ± 12
Male sex, %	70 (7)
White race, %	50 (5)
Comorbidities (%)
Diabetes	30 (3)
Hypertension	90 (9)
Heart failure	10 (1)
Physical examination
Weight, lb	251 ± 72
SBP, mmHg	144 (132–156)
Peritoneal Dialysis Variables
PD vintage, years	1.3 ± 0.9
APD utilization, %	100 (10)
Icodextrin use, %	30 (3)
Last fill, %	40 (4)
Etiology of Renal Disease
Diabetic nephropathy	2 (20%)
Hypertensive nephrosclerosis	1 (10%)*
Excess NSAID use
Systemic lupus erythematosus
Nephrotic syndrome
Immunoglobulin A nephropathy
Granulomatosis with polyangiitis-ESRD
Polycystic kidney disease Failed allograft
Medications (%)
Antihypertensives	90 (9)
Loop diuretics	60 (6)
Calcium channel blockers	60 (6)
Beta blockers	50 (5)
ACE inhibitors	30 (3)
Thiazide-type diuretics	20 (2)
Angiotensin II receptor blockers	20 (2)
Alpha agonists	10 (1)
Insulin	20 (2)
Laboratories
Sodium, mmol/L	137.9 ± 3.5
Hemoglobin, g/dL	10.2 ± 1.3
BUN, mmol/L	54 ± 19
Calcium, mmol/L	1.11 ± 0.14
Potassium, mmol/L	4.2 ± 0.4

^*= each of the etiologies listed, aside from diabetic nephropathy, account for 1 patient or 10% of the sample. SBP, systolic blood pressure; PD: peritoneal dialysis; APD: automated peritoneal dialysis; NSAID: non-steroidal anti-inflammatory drug; ESRD: end-stage renal disease; BUN: Blood urea nitrogen; ACE: angiotensin-converting enzyme

Figure 5:

Plasma glucose and glucose-corrected plasma sodium levels for individual human participants and aggregate data over a 2-hour dwell with direct sodium removal (DSR) or standard peritoneal dialysis (PD) solution.

Plasma glucose was higher with DSR at all timepoints during the dwell compared to standard PD solution (F(4;76)=5.3, p=0.008). Glucose corrected plasma sodium was not different between groups (F(4;76)=1.07, p=0.38). DSR: direct sodium removal, PD: peritoneal dialysis.

The secondary efficacy outcome of superior sodium removal with sodium free 10% dextrose (4.5 ± 0.4g) compared to standard PD solution (1.0 ± 0.3g) was met (p<0.001, Figure 6). In addition to a substantially higher average sodium clearance, the consistency of sodium removal was excellent (Figure 6). The absolute variability between individual patients’ sodium removal and the average sodium removal was similar between DSR and standard PD solution (Figure 6). However, since the total sodium removed in standard PD solution was substantially lower, the relative variability between individuals was much higher with standard PD solution a range of 8% to 75% of the average total sodium removal, compared with a range of 2% to 18% in DSR patients (Figure 6). Fluid removal was also greater with sodium free 10% dextrose, also with a high degree of consistence across patients.

Figure 6:

Volume of ultrafiltration and sodium removal in human subjects during a 2-hour dwell with DSR or standard PD solution.

Ultrafiltration volume (top left) and sodium content of peritoneal fluid (top right) over time in the aggregate population determined using I-131 indicator dilution. Ultrafiltration volume (bottom left) and sodium removal (bottom right) in individual patients at two hours determined from the physically drained volume. F(1;9) test statistics were =158 for ultrafiltration and 694 for sodium removal. DSR: direct sodium removal, PD: peritoneal dialysis.

Discussion:

The primary finding from this series of experiments is that substantial sodium removal via the peritoneal membrane is feasible. We found that using one liter of a sodium free 10% dextrose solution, which leverages both diffusive and convective forces, can remove over 4 grams of sodium in 2 hours both in healthy animals and humans. This therapy was well tolerated with limited effect on plasma electrolyte levels, minimal off target solute removal, and freedom from discomfort in the majority of human participants. The sodium removal was scalable, with substantially larger quantities of sodium removed by increasing the volume of 10% dextrose cycled into the peritoneal space. Furthermore, in the setting of experimental elevated right sided filling pressures, the rate and quantity of sodium removal was greatly increased. In aggregate, these data suggest that therapeutic strategies that directly target sodium removal across the peritoneal membrane may be valuable in HF and thus warrant additional study.

Loop diuretics represent the cornerstone of therapy for volume overload in HF. However, some degree of resistance to these agents is nearly ubiquitous. In the ROSE-AHF trial where all patients received high dose loop diuretics (per protocol 2.5 times home dose in 2 divided doses) study patients received a median 200mg of furosemide equivalents in the first 24 hours resulting in a median sodium output of 3.6 grams (interquartile range 1.9 to 6.0).²⁸ A two-hour dwell of sodium free 10% dextrose removed over 4 grams of sodium both in healthy normal swine and in patients receiving PD. Significantly more sodium was removed in animals with experimental HF and elevated right sided filling pressures. Potentially of equal or greater importance, the consistency of sodium removal was remarkably high with DSR, whereas diuretic response on an individual patient level is known to be highly variable.²⁴ Notably, in the ROSE-AHF trial, despite high dose intravenous diuretic administration, 29% of patients had a positive sodium balance, a finding which was strongly associated with worsened 6 month survival.²⁹ As such, given the large quantity and consistency of peritoneal sodium removal, there may be substantial advantage over traditional diuretic therapy in patients with diuretic resistance.

An additional potential advantage to direct sodium removal across the peritoneal membrane is an improved profile of off target solute removal. When a diuretic is given, invariably significant alterations in the tubular handling of non-sodium solutes occur. For example, the median potassium loss with a loop diuretic is 20mmol (interquartile range 12–31 mmol) and the average sodium output ~80 mmol.³⁰ As such the sodium-to-potassium ratio removed from the body is ~4:1.³⁰ This ratio only worsens with diuretic resistance and the use of adjuvant thiazide diuretics.³⁰ However, in the patients treated with 10% dextrose the ratio of sodium-to-potassium removed was 33:1 and this sodium and potassium removal ratio was remarkably similar across patients. The low removal of off target solutes is a simple matter of the diffusion gradient from plasma to the peritoneal solution. As opposed to sodium with a concentration >135mmol/L in most patients, the concentration of potassium, magnesium, and calcium are generally less than 5 mmol/L. As a result, the driving force for diffusion is nearly two orders of magnitude greater for sodium removal than off target electrolyte removal and thus high efficiency of sodium to off target solute removal is to be expected.

The regulation of water and sodium homeostasis have evolved in the following way: 1) The total quantity of water in the extracellular space is regulated by the retention/excretion of sodium and 2) the concentration of sodium in that water is regulated by fine tuning electrolyte free water consumption and excretion.⁵ Although dysregulation of sodium homeostasis is an early lesion in HF, the majority of patients with HF have an intact ability to regulate the concentration of sodium in the blood (i.e. an intact ability to excrete free water). Evidence for this is provided by the fact that the vast majority of HF patients have normal or near normal serum sodium.²⁹ This is despite having an expanded total quantity of both water and sodium in the body while at the same time chronically consuming a quantity of free water stoichiometrically greater than the limit one must remain under to prevent progressive hyponatremia (generally less than ~800 ml/day depending on urine concentrating ability).¹⁴ Rather, when HF patients, nearly ubiquitously, consume more fluid than this, they simply excrete the free water to maintain serum sodium levels, either normal or a modestly lower setpoint. In the current study we found that a two-hour dwell with sodium free 10% dextrose resulted in the “creation” of approximately one liter of free water by preferentially removing sodium over water from the body. Given the above described physiology of water handling in HF, the expectation is that in volume overloaded patients with acceptable renal function, free water will be excreted by the kidney following sodium removal preventing the complication of sodium removal. However, this hypothesis needs to be formally tested.

There has been interest in non-renal sodium and fluid removal dating back to before the 19th century with the use of Southey tubes. Contemporary non-renal fluid/sodium removal primarily leverages convective forces for sodium removal largely via trans-peritoneal ultrafiltration or veno-venous ultrafiltration. Most modern dialysis and ultrafiltration machines can modulate sodium/fluid removal and clearance of uremic solutes independently, allowing large quantities of fluid/sodium to be removed. However, a major limitation to veno-venous ultrafiltration as a chronic therapy is the required vascular access that can provide the >20 liters of blood per day required by most devices. This is usually accomplished through central venous catheters or grafts/fistulas leading to infection risk or a hemodynamic load respectively. Furthermore, veno-venous ultrafiltration systems are not currently available allowing patients to be fully ambulatory during treatment. In addition to the infection risk associated with a peritoneal catheter, in conventional PD the quantity of sodium that can be removed is limited by the fact that the solutions are designed to clear both uremic solutes and sodium. In order to allow large volume of solution to be cycled across the peritoneum for uremic solute clearance (generally 8–10L/day) sodium concentrations in these solutions are similar to that of plasma. However, since uremic solute clearance is not an objective of DSR, this approach can leverage both convective and diffusive forces, offering dramatic improvement in the efficiency of sodium removal with respect to both dwell time and required intraperitoneal volume.

Given the approximately 10-fold reduction in the volume of peritoneal solution required with DSR compared to standard PD, new approaches to movement of solution into and out of the abdomen are possible. The utility of DSR will likely be in chronic volume maintenance as secure access to the peritoneal cavity is not a trivial procedure. One potential fully implanted approach for a chronic therapy is use of a subcutaneous access port to the peritoneal space combined with a pump that transfers fluid from the peritoneum to the bladder (alfaPump, Sequana Medical). The alfaPump is a fully implantable, programmable, transcutaneous chargeable pump originally developed for the treatment of refractory and malignant ascites.³¹ Using this configuration, DSR solution could be introduced through the subcutaneous port. Then after a programmable delay for dwell of the solution, the pump could transfer the sodium rich fluid from the peritoneum to the bladder for removal from the body by urination. Currently, a study is being designed to test the safety and efficacy of the above configuration coupled with serial dosing of zero sodium peritoneal solution in HF patients.

Limitations:

The current experiments were designed to develop and provide human proof of concept for the direct peritoneal sodium removal concept. As such, these studies were not designed to provide a new indication for intra-peritoneal instillation of 10% dextrose solution for human use and do not inform the safety and efficacy of repeated dosing of a sodium free PD. Furthermore, the current study only demonstrated that DSR was symptomatically well tolerated, and acute asymptomatic peritoneal toxicity cannot be excluded. Only through chronic study can the long-term effects on plasma electrolytes and potential toxicity to the peritoneal membrane be determined. The physiology of an acute animal model created by a combination of tamponade and acute intravenous fluid loading is certainly not the same as a decompensated HF patient. As a result, based on these experiments we would hypothesize greater fluid/volume removal in humans with HF, these experiments should primarily be viewed as hypothesis generating. Study of chronic PD patients is a limitation since chronic changes in the peritoneal membrane may be different than HF patients. Furthermore, the I-131 albumin dilution method is not validated in PD patients and albumin leakage from the peritoneal membrane may confound these measurements. Additionally, the target population for DSR will ultimately be chronic HF patients with adequately functioning kidneys rather than ESRD patients. As such, the renal response to peritoneal sodium removal in a HF patient remains a major unknown (i.e., effects on GFR, sodium avidity and neurohormonal activation) and will need to be understood in future studies.

Conclusion:

DSR was well-tolerated in both animals and human subjects and produced substantially greater sodium removal than achievable with standard PD solutions. In animals with experimentally induced elevated right-sided filling pressures, sodium removal was further increased. Additional research evaluating the use of DSR as a method to prevent and treat hypervolemia in HF is warranted.

Clinical Perspective:

What is new?

• Cycling a sodium free osmotic solution, 10% dextrose, across the peritoneal cavity of swine resulted in substantial sodium removal.

• Sodium removal increased proportionately as the volume of 10% dextrose cycled across the peritoneum increased.

• Experimental elevation of right sided cardiac filling pressures also resulted in substantially increased sodium removal with this technique.

• A single dose of sodium free 10% dextrose was well tolerated in human subjects and resulted in over 4-fold greater sodium removal than the strongest commercially available peritoneal dialysis solution.

What are the clinical implications?

• Direct sodium removal with a sodium free osmotic peritoneal solution represents a new potential therapy for non-renal sodium and fluid removal in edematous disorders such as heart failure.

• Additional research to understand the safety and efficacy of direct sodium removal as a chronic therapy is warranted.

Non-standard Abbreviations and Acronyms

HF

Heart failure

DSR

Direct sodium removal

ESRD

End stage renal disease

PD

Peritoneal dialysis

NaCl

Sodium chloride

UF

Ultrafiltration

IQR

Interquartile range

ROSE-AHF

Renal Optimization Strategies Evaluation in Acute Heart Failure

APD

Automated peritoneal dialysis

CO

Cardiac output

MAP

Mean arterial pressure

PCWP

Pulmonary capillary wedge pressure

CVP

Central venous pressure