Conventional dialysis necessitates high volumes of dialysis fluid to maximize the concentration gradient between dialysate and blood side and enable an adequate removal of uremic toxins. In hemodialysis, between 100 and 180 L are used for every treatment—drinking water that has to be cleaned by passing through a complex water treatment system. Dialysate regeneration can save fluid, decrease the exposure of the patient to dialysate contaminants, and render water treatment systems unnecessary.
Sorbent technology was developed in the 1960s by Marquardt Corporation, an aerospace technology company in California, which was given the task of inventing a system for astronauts to generate drinking water from waste. Sorbents can employ different types of interactions: cation exchange, anion exchange, van der Waals interactions, and biologic binding. The affinity of organic compounds to activated charcoal and nonionic macroporous resins is due to van der Waals forces and depends on the number of pores and pore size.
Sorbent technology in hemodialysis was introduced in the 1970s. With the REDY machine (Recirculating Dialysis, Organon Teknika, Oss, The Netherlands), spent dialysate was cleared of uremic toxins while passing through a charcoal-containing cartridge and reused subsequently.1 In contrast to single-pass hemodialysis, only 6 L of tap water was required. Further refinement of cartridge and machine has led to the development of the Allient dialysis system (Renal Solutions Inc., Warrendale, PA).2 The pursuit of cost reductions, the increasing appeal of home hemodialysis, and the prospect of the long-awaited wearable artificial kidney have continued to spur interest in sorbent technology.3
A fluid quality equal to that seen in conventional dialysis must be the objective for long-term use of sorbent-regenerated dialysate in chronic dialysis patients. One aspect is bacteriological purity of the dialysate, and the other prominent domain of dialysis quality relates to solute removal efficiency. Treatment quality comparable to single-pass dialysis can only be achieved provided the near-complete removal of uremic toxins before the reuse of the fluid.
Three broad categories of uremic retention solutes can be identified: 1) small, water-soluble compounds, the prototypes being urea and creatinine, 2) the so-called “middle molecules,” and 3) protein-bound solutes, for example p-cresol sulfate and indoxyl sulfate. While the removal of small, water-soluble molecules is well documented with the use of commercially available sorbent cartridges, data concerning sorbent efficiency for the other two classes are scarce. Prominent representatives from each of these categories are beta-2- microglobulin (β2M) (molecular weight: 11818 Da) and p-cresol sulfate (188 Da, 93%–96% protein bound4,5)—solutes for which uremic toxicity has unequivocally been demonstrated.
p-Cresol is a product of the metabolism of phenylalanine and tyrosine by intestinal anaerobic bacteria. Most of the p-cresol generated is conjugated to p-cresol sulfate (PCS) in the intestinal wall and to p-cresol glucuronide in the liver. PCS serum concentration is strongly associated with cardiovascular morbidity and mortality in hemodialysis patients6 and has been shown to activate free radical production by leukocytes. Plasma PCS levels are linked to circulating levels of endothelial microparticles (EM) in vivo, and PCS induces shedding of EM in vitro.7 EM impair the NO signaling pathway, act prothrombotically and proinflammatorily, and lead to endothelial dysfunction. Beta-2-microglobulin is strongly associated with morbidity and mortality in dialysis patients.8,9
Recently, the regeneration of peritoneal dialysate has emerged as one focus in the development of a wearable artificial kidney.3,10 The regeneration of outflow peritoneal dialysate, however, represents a somewhat different situation from hemodialysis in that peritoneal dialysate contains significant amounts of albumin. It is desirable to prevent a contact between albumin and the sorbent to avoid impairment of the urease layer. This can be accomplished by dialyzing spent peritoneal dialysis (PD) dialysate against a secondary dialysate stream, which in turn is passed through the sorbent rather than passing the spent dialysate directly through the sorbent cartridge.
In this study, we characterized the efficiency and capacity of sorbent technology for clearance of PCS from the dialysate. We further investigated the feasibility of clearing spent peritoneal dialysate of β2M using a high-flux dialyzer and a countercurrent sorbent circuit.
The study was approved by the Beth Israel Medical Center Institutional Review Board, and all participating subjects gave written informed consent before enrollment.
The cartridges used were manufactured by SORB Technology, Inc., Oklahoma City, OK. They consisted of four different layers: 1) activated carbon (224 g) for removal of heavy metals, organic components, oxidants, chloramine, creatinine, uric acid, and middle molecules; 2) vegetable-derived immobilized urease (205 g) for enzymatic degradation of urea, which results in release of NH4+ and HCO3−; 3) zirconium phosphate (700 g) that binds potassium, calcium, magnesium, and other cations in exchange for Na+ and H+; and 4) zirconium oxide and -carbonate (80 g each) that exchange anions for acetate, HCO3−, and Na+. A new sorbent cartridge was used for each experiment.
PCS was synthesized following the protocol of Feigenbaum and Neuberg.11 One gram of p-cresol was dissolved in 5 ml of pyridine, followed by dropwise addition of 1 ml chlorosulfonic acid, while stirring on ice. The product was neutralized with addition of 10 ml 3N potassium hydroxide and washed with ether in a Buchner funnel with filter paper. After washing with ethanol twice, 0.7 g of PCS was obtained. Product formation was confirmed by high-pressure liquid chromatography (HPLC) analysis. Based on HPLC analysis, PCS purity was estimated to be >90%.
PCS synthesis was validated by conversion of the product to p-cresol via digestion with Helix Pomatia sulfatase for 12 hours in pH 5 sodium acetate buffer.
PCS concentrations were analyzed on an Agilent 1100 system HPLC using a 25 cm × 0.46 cm Beckman C18 Adsorbosphere column; the mobile phase was 65% 0.1 M ammonium formate, pH 5, and 35% methanol, at 1 ml/min flow rate. Fluorescence detection was accomplished with excitation at 214 nm and emission at 310 nm. Detection limit with the HPLC was 10 ng, and 50 μL aliquots of the samples were injected on the column.
PCS Removal Studies
Short Clearance Studies.
For a PCS concentration of 10 mg/L, 40 mg of PCS powder was added to 4 L of fresh dialysate (pH 7.4) containing sodium 130 mmol/L, chloride 102 mmol/L, calcium 1.25 mmol/L, magnesium 0.5 mmol/L, glucose 100 mg/dl, and bicarbonate 30 mmol/L. The experimental setup is depicted in Figure 1A. After priming the cartridge with water, the PCS solution was recirculated through the cartridge for 1 hour at a flow rate of 250 ml/min. The dialysate reservoir was stirred for adequate mixing. Samples were collected upstream (“pre”) and downstream (“post”) of the sorbent cartridge every 10 minutes and stored at −70°C until HPLC analysis. Five experiments were conducted.
Employing the same setup as above, an additional experiment was conducted using a recirculation time of 3 hours and a solution of 240 mg of PCS powder in 12 L of dialysate, resulting in a threefold increase in sorbent perfusion time, twice the concentration (20 mg/L PCS), and six times the total amount of PCS as was used in the short studies. This approximates the amount of PCS cleared during a typical high-flux hemodialysis treatment.4 The purpose of this experiment was to uncover a potential decrease in cartridge performance or a limit in the adsorption capacity with clinically relevant PCS loads.
Beta-2-Microglobulin Removal Studies
Four batches of spent peritoneal dialysate (13.6, 14.6, 25, and 22.7 L, respectively) were collected from patients on PD. Each batch was concentrated down to 1 L using pressure filtration through a low-flux dialyzer membrane (F4, Fresenius Medical Care NA, Waltham, MA). Each concentrate was studied as follows (Figure 1B). The spent PD fluid was recirculated on the “blood” side of a polysulfone high-flux dialyzer (F160NR, Fresenius Medical Care NA, Waltham, MA) for 1 hour (“PD fluid circuit”). On the dialysate side of the membrane, in countercurrent flow, 4 L of bicarbonate dialysate as characterized above was recirculated through a sorbent cartridge (“sorbent circuit”). PD fluid and sorbent circuits were maintained at a flow rate of 250 ml/min. Fluid samples for measurement of β2M were collected every 10 minutes upstream (“pre”) and downstream (“post”) of the dialyzer in the PD fluid circuit as well as upstream (“pre”) and downstream (“post”) of the cartridge in the sorbent circuit. β2M was measured turbidimetrically using an Olympus analyzer and β2M reagent containing latex particles coated with anti-β2M IgG antibodies [Olympus Diagnostica GmbH (Irish Branch), Lismeehan, O' Callaghan's Mills, Co. Clare, Ireland].12
The β2M clearances for each experiment were calculated for each available time point by calculating the instantaneous flux across the blood compartment and dividing it by the instantaneous β2M concentration gradient, where the instantaneous flux was calculated as
with Jβ2M being the instantaneous β2M flux, Qe being the flow rate through the blood compartment of the dialyzer, and Cbo and Cbi being the blood outlet and inlet concentrations, respectively, of β2M, and the concentration gradient was calculated as dialysate inlet minus blood inlet β2M concentration. As postcartridge β2M levels were consistently below the detection limit, a value of zero was assumed for dialysate inlet concentration in these calculations.
For PCS, the clearances for each experiment were calculated for each available time point as
with KPCS being the instantaneous PCS clearance, t being the elapsed time after starting the experiment, C0 being the precartridge PCS concentration at the beginning of the experiment, Ct being the precartridge PCS concentration at time point t, and V being the distribution volume of PCS in the respective experiment.
The clearances obtained as described above for β2M and PCS are referred to as “calculated clearances” in this article. Data points with blood inlet concentrations below the detection limit were excluded from clearance calculations.
Clearances were further estimated by fitting one-phase exponential decay functions through the concentration time course data, thereby obtaining the exponential decay rate constant k and calculating the clearance K as
with V being the distribution volume of β2M or PCS in the respective experiment.
Determination of β2M Mass Balance Closure
β2M mass balance across the blood compartment of the dialyzer was determined by deriving the rate of solute transfer from solute mass entering and leaving the blood compartment and fitting a one-phase exponential decay function to the data points. This function was then integrated and used to derive the cumulative solute mass transfer across the membrane. Similarly, sorbent circuit mass balance was derived by calculating the rate of β2M appearance on the sorbent circuit side of the membrane, fitting a one-phase exponential decay function to the data, and integrating the function to derive the cumulative β2M mass transfer into the sorbent circuit.
Calculation of the Mass Transfer Area Coefficient for β2M
The dialyzer-specific mass transfer area coefficient (KoA) for β2M was calculated as
with K being the solute clearance, Qe being the solute diffusion volume flow rate on the blood side of the dialyzer, and Qd being the dialysate flow rate on the dialysate side of the dialyzer.
Data are presented as mean ± SD unless otherwise noted. Error propagation was taken into account when determining clearances or KoA across multiple experiments. The PCS clearances between the short PCS removal studies and the PCS exhaustion study were compared by standard equivalency test using 1.96 SD (i.e., 95% confidence) as the breakpoint for agreement.
PCS Removal: Short Clearance Studies
The initial PCS concentration in the experimental dialysate was on average 0.88 mg/dl. Over the course of the five 1-hour experiments, PCS was virtually completely removed from the dialysate (Figure 2). On average, the relative precartridge PCS concentration in the fluid was 52.4%, 25.0%, 13.6%, 7.0%, 3.0%, and 1.4% after 10, 20, 30, 40, 50, and 60 minutes, respectively, compared with the starting concentration. The calculated PCS clearance across all five experiments was 275.7 ± 19.7 ml/min. Nonlinear curve fitting revealed a decay rate constant of 0.06538 min−1 [95% confidence interval (CI): 0.04224–0.08852]. This translates into a clearance of 261.5 ml/min, which is in agreement with the calculated clearance. Postcartridge concentrations were usually undetectable but always ≤0.05 mg/dl, indicating nearly complete extraction of PCS across the sorbent cartridge, even at PCS concentrations of around 1 mg/dl, which is about 10-fold higher than would be expected on average to appear in spent dialysate during single-pass hemodialysis.4
PCS Removal: Sorbent Capacity Study
The starting concentration of PCS for this experiment was 2.16 mg/dl, corresponding to a total amount of 259 mg PCS. Over the course of the 3-hour experiment, PCS was nearly completely cleared from the dialysate (Figure 3). The relative precartridge PCS concentration in the fluid mix was 81.5%, 53.2%, 37.0%, 14.8%, 7.9%, 4.3%, and 2.4% after 15, 30, 60, 90, 120, 150, and 180 minutes, respectively, compared with the starting concentration. The calculated PCS clearance for this experiment was 232.2 ± 36.3 ml/min. Curve-fitting of the data showed a decay rate constant of 0.01790 min−1 (95% CI: 0.01198–0.02383), with a corresponding clearance of 214.8 ml/min. Postcartridge PCS was always undetectable, i.e., the PCS extraction fraction across the sorbent cartridge remained close to 100% throughout the entire 3 hours without any performance drop. There was no statistically significant difference in PCS clearance between this experiment and the short clearance studies.
Beta-2-Microglobulin Removal Studies
Pressure filtration resulted in final β2M starting concentrations between 8.1 and 27.5 mg/L. The average extraction fraction of β2M across the blood compartment of the dialyzer was 12.8% ± 4.9%. The highest resulting precartridge β2M concentrations observed were between 0.5 and 1.8 mg/L. Postcartridge concentrations were consistently below the validated range of the test (lower limit 0.5 mg/L) and on average <0.1 mg/L. The relative predialyzer concentration of β2M in the PD fluid was on average 62.1%, 48.3%, 36.8%, 28.6%, 20.4%, and 14.3% after 10, 20, 30, 40, 50, and 60 minutes, respectively, relative to the starting concentration. Figure 4 shows the time course of predialzyer β2M concentration in the peritoneal dialysate for all experiments, normalized to the respective starting concentrations. The calculated clearance of β2M across all four experiments was 32.4 ± 6.3 ml/min. Curve-fitting yielded a decay rate constant of 0.06538 min−1 (95% CI: 0.04224–0.08852), which translated into a corresponding clearance of 43.3 ml/min. The β2M KoA for the F160 dialyzer was found to be 38.3 ± 9.3 ml/min. Cumulative β2M loss across the blood compartment of the dialyzer in the experiment with the highest starting concentration totaled 22 mg. The total β2M mass appearing in front of the sorbent cartridge was 18.9 mg, i.e., 86% of the total mass removed from the PD fluid.
The presented results provide proof of concept that dialysate can be cleared of PCS by passing it through a commercially available charcoal-containing sorbent cartridge. Moreover, two results are of particular importance for the use of regenerated fluid in chronic dialysis. First, virtually all PCS was cleared during a single passage through the sorbent. Of note, the PCS concentrations presented to the cartridge at the beginning of the experiments were much higher (by a factor of >8 on average for the short clearance studies and by a factor of >20 for the capacity study) than is encountered in spent dialysate during regular hemodialysis treatments (∼0.1 mg/dl).4 Consequently, the rate of PCS presentation to the sorbent was markedly higher than the rate of PCS clearance across the dialyzer which would be expected to occur in high-flux dialysis (up to 5.4 mg/min compared with ∼0.9 mg/min).4 This translates into practically PCS-free dialysate after only a single passage, which can be reused without concerns of a reduction in PCS clearance. Second, neither cartridge exhaustion nor PCS leakage did occur in the course of the exhaustion experiment despite an adsorbed mass of PCS totaling >250 mg. The targeted amount of approximately 240 mg PCS in the initial dialysate was chosen so as to exceed the range of PCS removed with the dialysate during a regular hemodialysis treatment, usually 100–220 mg.4,13 The kinetics of PCS removal were not different between the short studies and the exhaustion experiment. We conclude that a single sorbent cartridge has sufficient capacity to fully clear PCS from spent dialysate over the course of a regular hemodialysis session. However, it is important to stress that regeneration of dialysate per se can never increase the removal of any solute beyond what is seen with regular single-pass hemodialysis, as it can at best be equivalent to fresh dialysate. An increase in removal can only be achieved by changing the dialyzer membrane properties, increasing dialysate and blood flow rates, or by suspending a sorbent directly in the dialysate. The free concentrations of protein-bound retention solutes in plasma are low, resulting in rapid annihilation of the concentration gradient along the dialyzer. Employing a sorbent suspension (such as carbon slurry) as dialysate ideally allows maintenance of the concentration gradient along the entire length of the dialyzer, the effect being equivalent to an infinitely high dialysate flow rate during single-pass hemodialysis.14
Activated charcoal is a common, but not the only, sorbent used today. Meijers et al.15 were able to show that PCS is also removed by a resin adsorbent combined with ion exchange as used in the Prometheus system for treatment in liver failure.
The second group of experiments presented here demonstrates that PD fluid can be cleared of β2M using a high-flux dialyzer and a countercurrent sorbent cartridge circuit.
While direct contact of the blood or peritoneal dialysate with the sorbent would provide a superior removal of β2M, the setup used here provides some major advantages. Hemoperfusion across sorbents was considered as an alternative to hemodialysis early on in the evolution of renal replacement therapy.16,17 Contact of blood with charcoal, however, caused destruction of platelets, complement activation, and sorbent embolization. In the setting of PD, albumin extraction, as occurs when passing peritoneal dialysate through a sorbent directly, is undesirable. Also, activated charcoal does not take up urea and is, therefore, combined with urease in modern sorbent technology. The performance of the urease layer, however, is negatively affected by albumin. For these reasons, separation of the sorbent circuit from the blood or PD dialysate stream by a high-flux dialyzer membrane is a favorable option. A pertinent question in this setting, however, is to what extent β2M is adsorbed to the membrane of the dialyzer and what fraction of the total removal can be attributed to the sorbent. The presented mass balance calculations demonstrate that the bulk (86%) of β2M leaving the blood side circulation in fact appears in the sorbent circuit and, hence, is presented to the sorbent cartridge. Furthermore, the KoA for β2M that we found (38.3 ± 9.3 ml/min) is in line with data reported in the literature. Bhimani et al. recently reported a KoA for β2M for the Fresenius Optiflux F160NR dialyzer of 27 ± 2 ml/min at a blood flow rate of 400 ml/min and a dialysate flow rate of 350 ml/min.18 The lower KoA in the study by Bhimani et al. is not unexpected, considering that they measured β2M clearance in vivo, with red blood cells acting as a diffusion barrier to β2M, whereas we measured β2M clearance in peritoneal dialysate.
Of note, the purpose of our study was not to determine the β2M binding of carbon but rather to study the application of sorbent technology for regeneration of peritoneal dialysate. This goal dictated our choice of study design. We deliberately studied the entire sorbent cartridge as a unit as opposed to studying β2M removal using an isolated carbon column, as the results may not necessarily be identical. The two circuits were used to protein-deplete the PD fluid before passage through the sorbent, which is necessary to prevent damage to the urease layer. Also, other layers in the sorbent cartridge (the urease layer, which uses alumina to immobilize the urease enzyme, and the zirconium oxide/zirconium carbonate anion layer) are expected to have affinity for β2M and, although positioned after the carbon layer in the flow stream, may or may not contribute to β2M removal in practice. Finally, running native PD fluid through an isolated carbon cartridge may not necessarily produce the same results as using albumin-depleted PD fluid. Although carbon has the characteristic of adsorption sites being somewhat molecular weight-specific, it also has some electrostatic adsorption characteristics that are not as specific regarding proteins, so the presence of large quantities of other proteins may or may not influence the binding of β2M.
Currently, several groups are developing prototypes of wearable artificial kidneys. Some of these use a hemodialysis and others a PD-based approach, but because carrying replacement fluid is not a viable option in this setting, all of them are working with sorbent-regenerated dialysate.3,10,19,20
Spent PD fluid can be cleared of β2M by dialyzing it across a high-flux membrane against a countercurrent dialysate circuit containing a commercially available sorbent cartridge. p-Cresol sulfate in amounts typically removed during conventional single-pass hemodialysis is nearly completely cleared from dialysate using the same sorbent technology.