Secondary Logo

Journal Logo

Kidney Support

Sorbent Hemodialysis: Clinical Experience With New Sorbent Cartridges and Hemodialyzers

McGill, Rita L.; Bakos, Jane R.; Ko, Tina; Sandroni, Stephen E.; Marcus, Richard J.

Author Information
doi: 10.1097/MAT.0b013e318185e66a
  • Free


Sorbent hemodialysis was an early variant of renal therapy, in which spent dialysate was reprocessed and used repeatedly.1–5 Sorbent hemodialysis required only 6 liters of tap water, in contrast to 30–50 L/h required for conventional (single-pass) hemodialysis.5,6 Sorbent machines reclaim spent dialysate in two steps. Hemodialyzer effluent first passes through a purification cartridge (Figure 1) which removes organic molecules, uremic toxins, and inorganic impurities. Urea, which is difficult to adsorb, must be enzymatically converted to ammonium and removed in subsequent layers of the cartridge. Dialysate regeneration is completed by the addition of a concentrated electrolyte solution which restores potassium, calcium, and magnesium.1,2,4,5

Figure 1.
Figure 1.:
Composition of a modern sorbent dialysate regeneration cartridge. Spent dialysate enters the bottom of the cartridge and exits at the top. The first layer, an activated carbon layer, removes heavy metals, organic compounds, and middle molecules. The next layer, a vegetable-derived urease fixed to an inorganic matrix, enzymatically degrades urea to ammonium carbonate. The third layer, a cation exchanger, takes up ammonium and inorganic cations and emits sodium and hydrogen. The fourth layer, an anion exchanger, takes up phosphates, fluoride and heavy metals, releasing sodium, bicarbonate, and small amounts of acetate.

Sorbent hemodialysis evolved from water recycling technology originally intended for astronauts and reached fruition as the REcirculating DYalysis (REDY) Sorbent System, which was marketed from 1973 to 1993.4 Sorbent machines have performed thousands of treatments3 and are still in limited use, although manufacture of the REDY system ceased in 1993. The resource allocation of sorbent dialysis differs from single-pass hemodialysis, and the evolution of hemodialysis into centralized outpatient facilities favored the economics of bulk water purification, which permitted the expansion of single-pass methodology.7

Residual use of the REDY system today occurs where water supply is an issue: arid areas, remote areas, and military applications. REDY was useful in the medical relief response to the 1998 Armenia earthquake, when disaster conditions precluded water generation for conventional single-pass hemodialyzers.8

The superior results with extended length and nocturnal hemodialysis have led nephrologists to think beyond the conventional dialysis center model. The feasibility of sorbent dialysis changes when home hemodialysis is considered. The water requirements can be a serious obstacle to many home budgets and local water systems, without considering the additional water volumes rejected from a reverse-osmosis system. Prepackaged dialysates create storage space and weight issues.

Hemodialysis has improved in many ways over the past 25 years, with the introduction of ultrafiltration control, bicarbonate dialysates, biocompatible membranes, and faster blood pumps. The REDY, which went out of production in the early 1990s, incorporates none of these features. Re-engineering of sorbent hemodialysis resulted in a new machine, the Allient Hemodialysis System (Renal Solutions, Inc., Warrendale, PA).6,9 The modern machine has ultrafiltration control and can safely accommodate modern high-flux hemodialysis membranes.

Larger sorbent cartridges have been developed, which can generate dialysates for up to 8 hours. In the new cartridges, the terminal activated charcoal layer has been incorporated into the initial purification layer, resulting in four layers, rather than five. The original anion exchange layer is now composed of zirconium oxide and zirconium carbonate, rather than hydrous zirconium oxide, to favor the generation of bicarbonate over acetate.

First, an activated carbon layer removes heavy metals, oxidants, chloramines, and other organic molecules. As direct adsorption of urea is not feasible, the next layer converts urea to ammonium using the enzyme urease. Ammonium released by the urease layer is bound by the third layer, composed of zirconium phosphate, which also eliminates calcium, magnesium, potassium, and other metals from the dialysate. As cations are bound, sodium and hydrogen are released into the dialysate, in a ratio that depends upon the pH of the dialysate. A final layer of zirconium oxide and zirconium carbonate adsorbs phosphates, fluoride and heavy metals, while releasing bicarbonate, sodium, and small amounts of acetate. The dialysate leaving the cartridge is a near-ultrapure solution consisting of water, sodium chloride (NaCl), and alkali, as sodium bicarbonate (NaHCO3) or sodium acetate. Endotoxin testing shows the purity of sorbent dialysates exceeds both the 2001 AAMI (Association for the Advancement of Medical Instrumentation) water quality standards and the European Renal Association best practice guidelines.10

SORB + cartridges are designed to remove 9.5–23.5 g of urea over 3–6 hours of dialysis; HISORB+ cartridges can remove 23.5–35 g. SORB HD and HISORB HD cartridges operate over 6–8 hours. The SORB HD can remove 12.5–20 g, and the HISORB HD can remove 15.0–30 g of urea.

Features of the Allient that promote safer and simpler operation for home users include on-line air detection, multiple pressure transducers, and automatic priming and rinseback procedures. The Allient incorporates a continuous, on-line ammonia detection device, linked to an alarm system that can halt the machine.9 Although the enzymatic layer can convert urea indefinitely, the zirconium phosphate cation exchange can be exhausted, in which case, the cartridge releases ammonia into the dialysate.1–3 This innate risk of sorbent dialysis previously required cumbersome manual testing of the dialysate with strips of test paper.

The REDY, like most contemporaneous machines, had maximal blood and dialysate flow rates of only 250 ml/min. The capacity of the early sorbent cartridges could be prematurely exhausted at higher dialysate flow rates, but the higher capacity of the new cartridges permitted the Allient System to be designed for faster pump speeds. The blood pump is designed without rollers and can generate blood flow up to 400 ml/min. The dialysate pump generates flow rates of 200–400 ml/min.9 The Allient blood pumps have a pair of polycarbonate resin chambers integrated into the disposable bloodlines. Flexible plastic diaphragms, located on the medial aspects of both chambers, are displaced by pneumatic pressures generated by the Allient machine. The Allient System operates with a conventional double-needle or double-lumen vascular access, but can be adapted easily to work with a single vascular access port.9

An important consideration in using sorbent cartridges is the characteristic evolution of the dialysate composition over the course of a treatment. Sorbent cartridge action generates NaCl and NaHCO3, so dialysate becomes more alkaline and more hypertonic during normal operation. The initial dialysate is often intentionally hypotonic, to offset the sodium generated as the sorbent cartridge exchanges ammonium for sodium in the zirconium phosphate layer, which in turn depends upon the urea nitrogen burden of the individual patient. As a sorbent dialysis proceeds, sodium moves down a concentration gradient from patient to dialysate, a process that is mitigated by the total dialysate volume which is much smaller than the total body water of the patient. As the cartridge releases sodium, the dialysate becomes hypertonic to plasma, and concentration gradient ultimately favors transfer of sodium from dialysate to patient. As before, sodium transfer is mitigated by the relatively small volume of dialysate compared to the total body water. Although the dialysate frequently becomes hypertonic, the patient does not. The overall effects on sodium are net removal from the body, which has allowed sorbent hemodialysis to be used successfully in chronic outpatient dialysis.1–5 Concurrent volume removal often results in an increase of serum sodium by 1–2 mEq/L, similar to single-pass hemodialysis. However, the dialysate sodium profile is precisely the reverse of most single-pass hemodialyzer “sodium modeling” programs, in which increased dialysate tonicity early in treatments reduces uncomfortable dialysis symptoms.11

The physical and chemical behavior of the dialysate produced by the new cartridges was investigated extensively in vitro, to refine equations to describe the mass-transfer of sodium between a patient, a dialysate, and a sorbent column.12,13 These predictive models were incorporated into clinical software, which has evolved into a program called the Prescription Guide, designed to assist ordering physicians. Prescription Guide uses calculations derived from multiple linear regression analyses of relevant clinical parameters, including: extracellular volume, treatment duration, dialyzer clearance, ultrafiltration volume, replacement fluid volume, and predialysis lab values. The software provides predictions of the expected values for serum sodium and bicarbonate after therapy, as well as the predicted Kt/V. Prescription Guide was developed to assist clinicians selecting treatment parameters (e.g., initial dialysate composition, membrane, cartridge, treatment duration). The clinical data from this trial were used to evaluate the usefulness and predictive accuracy of this software, when the new cartridges were used for sorbent hemodialysis.


This trial was conducted in accordance with the guidelines of the International Conference on Harmonization for Good Clinical Practice. The study protocol and consent form were approved by the Institutional Review Board. Ambulatory patients were recruited from the outpatient hemodialysis unit located at Allegheny General Hospital. Informed consent was obtained from all subjects prior to enrollment. Exclusion criteria included age <18 years, duration of outpatient hemodialysis <4 weeks, unreliable vascular access, pregnancy, inability to pursue contraception, or lactation.

Charts were reviewed after the study to obtain an average Kt/V from three consecutive monthly values (the months before, during, and after the sorbent study treatment occurred). Treatment length and average blood flow rate were also recorded.

Hemodialysis orders were written by nephrologists trained in sorbent hemodialysis. Patients underwent hemodialysis with either a REDY or an Allient Hemodialysis System in dual-access mode, with dedicated nursing staff trained in the use of the machine at chair-side throughout treatments. Allient treatments were performed using Polyflux 10 L hemodialyzer membranes (Gambro, Lakewood, CO), rather than the larger Polyflux 210H membranes used during the regular single-pass dialysis treatments. A smaller CA-HP130 membrane (Baxter Corporation, Deerfield, IL) was used with the REDY, which lacked volumetric control of ultrafiltration. Hypotension and nausea were recorded. Most symptoms were treated with small boluses of intravenous saline; mild muscular cramping was treated with hypertonic mannitol, as well.

Blood samples were withdrawn from vascular access devices at the beginning and end of sorbent hemodialysis. Sorbent dialysate samples were withdrawn from the system at the outset of dialysis, at treatment mid-point, and at the end of dialysis. Dialysate conductivity and pH were tested with a hand-held meter every 15 minutes. Dialysate samples were tested for sodium, potassium, acetate, total CO2, urea nitrogen, creatinine, calcium, magnesium, phosphorus, and ammonia.

Blood samples were tested for sodium, potassium, chloride, total CO2, urea nitrogen, creatinine, glucose, calcium, phosphorus, magnesium, albumin, and a complete blood count was performed. These values were used to derive anion gap, single-pool Kt/V (using a modified Daugirdas14 equation), and urea nitrogen removal.

After the trial, treatment parameters, patient demographics, and laboratory data were entered into the Prescription Guide software. Software predictions of Kt/V were compared to results calculated from actual postdialysis labwork. Prediction values were derived from the labwork available to the ordering physician, consisting of labwork < 30-days old. Mean values for predicted Kt/V and actual Kt/V were examined using a paired Student’s t test. Pearson’s coefficient of correlation was also calculated.


During the study, 19 patients were scheduled for 31 treatments. One patient failed to show up for scheduled appointments, and six treatments were censored because prescribed treatments were not delivered due to various clinical and technical issues. Among 25 completed treatments, 10 were performed with the REDY, and 15 were performed with the Allient. Table 1 contains demographic data on the 19 patients, who resembled a typical hemodialysis outpatient population. Dialysis prescriptions are described in Table 2. Laboratory parameters before and after sorbent dialysis are listed in Table 3.

Table 1
Table 1:
Demographic Data of Therapy Calculator Patients
Table 2
Table 2:
Sorbent Hemodialysis Prescriptions
Table 3
Table 3:
Numerical Parameters Before and After Sorbent Dialysis

Sorbent treatment durations ranged from 3.5 to 6 h (mean = 4.86 h, compared to 4.05 h for single-pass treatments.) The average blood flow rate of sorbent treatments was 321 ml/min; largely because the REDY was limited to a peak blood flow of 250 ml/min. As regular single-pass treatments were conducted at 450–500 ml/min, processed blood volumes were considerably lower during sorbent therapy. REDY treatments rarely processed more than 80 liters of blood, compared to 100–120 liters for single-pass treatments. Allient treatments ranged from 80 to 130, dependent upon prescribed dialyzer time.

Kt/V during sorbent treatments ranged from 0.85 to 1.62, compared to 0.69–1.93 for the subjects’ regular outpatient dialyses. Mean Kt/V was lower in sorbent treatments (1.27 vs. 1.49, p = 0.0001).

Clinical issues were infrequent. Five patients experienced nausea or dizziness, none severe enough to warrant discontinuing treatment. Symptoms were slightly more frequent during REDY treatments, but did not reach statistical significance. Most symptoms occurred approximately 70–90 minutes into treatment, and lasted <5 minutes. No symptoms occurred beyond 130 minutes, aside from mild muscular cramping, even when treatments were prolonged. Symptomatic patients trended toward higher predialysis blood glucose (136 vs. 100 mg/dl), lower serum bicarbonate (21 vs. 42 mEq/L), and lower initial dialysate sodium (109 vs. 118 mEq/L), but none of these trends reached statistical significance. Ultrafiltration rates were similar. Mean Kt/V achieved was 1.27 ± 0.24, vs. a predicted value 1.35 ± 0.25 (p = 0.17, NS), although individual values were subject to variable degrees of scatter (Figure 2).

Figure 2.
Figure 2.:
Performance of the Prescription Guide software, showing predicted vs. actual Kt/V. R = 0.39, indicating a weak relationship.


This study provides a detailed clinical and biochemical exploration of the interaction between a modern sorbent dialysis system and an initial group of kidney patients. The data provide valuable new insights into the practice of sorbent hemodialysis.

The Prescription Guide software performed well as a predictor of posthemodialysis chemistries, which permitted study physicians to evaluate different dialysate strategies with ease and assurance. One surprising outcome was a distinct and progressive drift in initial dialysate sodium concentrations chosen by our physicians, who expressed an instinctive discomfort with “nonphysiological” sodium levels. Although initial treatments were performed with dialysate sodium levels of 100–110 mEq/L, in accordance with existing guidelines for sorbent hemodialysis, later treatments started with dialysate sodium levels of 114–136 mEq/L. Postdialysis sodium and bicarbonate levels continued to be acceptable even as initial dialysate sodium levels rose; hypernatremia was not observed. Volume removal of up to 6 liters was readily achievable, though rarely ordered; average volume removal was 3.3 liters.

Sorbent dialyses could achieve clearances that met clinically relevant goals, although, as in all forms of hemodialysis, slower blood flow rates could compromise clearance at shorter treatment times. Slow-flow sorbent treatments therefore must be prescribed with adequate dialysis time to ensure good clearances. Future study is merited to elucidate a comparison of clearances between sorbent and single-pass dialyzers operating under similar circumstances.

The Prescription Guide software provided a rough estimate of expected clearances, with considerable variability, some of which may have resulted from comparing actual clearances based on same-day labwork, to predictions based on less current lab values (a situation more representative of actual clinical practice.) Clinicians will need to measure urea clearances to guide the dialysis prescription.

Treatments were tolerated well, and the rate of symptoms not different from the same patients dialyzing on their regular single-pass machinery. Patients left the unit ambulatory and well, and most were enthusiastic about the striking absence of symptoms during the final hour of hemodialysis. This study was not designed to determine the reason for the unanticipated change in the time course of dialyzer symptomatology. Given the salutary effects of hypertonic NaCl and dialysate sodium modeling, we hypothesize that the rising dialysate sodium may have exerted a progressively stabilizing effect on symptoms. Further study is needed to determine whether “reversed” sodium modeling in dialysate actually reduces dialysis symptoms.


Modern sorbent hemodialysis can deliver extended treatments with adequate urea clearances and ultrafiltration, using only 6 liters of tap water. Using modern equipment, treatments were well-tolerated and late-treatment symptomatology was rare, even when treatments were prolonged. New sorbent machines present a promising alternative technology to bring high-quality hemodialysis into venues where water supply may be an issue and may play an important role in making home hemodialysis more accessible.


1. Gordon A, Better OS, Greenbaum MA, et al: Clinical maintenance hemodialysis with a sorbent-based, low-volume dialysate regeneration system. ASAIO Trans 17: 253–258, 1971.
2. Blagg CR, Vizzo JE, Jensen WB, Cole JJ: Experience with a sorbent-based dialysate regeneration system for hemodialysis. Prog Biochem Pharmacol 9: 239–248, 1974.
3. Blumenkrantz MJ, Gordon A, Roberts M, et al: Applications of the REDY sorbent system to hemodialysis and peritoneal dialysis. Artif Organs 3: 230–236, 1979.
4. Roberts M: The status of sorbent technology in hemodialysis treatment. Clin Nephrol 26(suppl 1): S44–S46, 1986.
5. Roberts M: The regenerative dialysis (REDY) sorbent system. Nephrology 4: 275–278, 1998.
6. Shapiro W: The allient® sorbent hemodialysis system, in Nissenson AR, Fine RN (eds), Handbook of Dialysis Therapy, 4th ed. Philadelphia, PA, Elsevier Mosby Saunders, 2007, pp. 512–520.
7. Hansen SK: Advances in sorbent dialysis. Dial Transplant 34: 648–652, 2005.
8. Richards NT, Tattersall J, McCann M, et al: Dialysis for acute renal failure due to crush injuries after the Armenian earthquake. BMJ 298: 443–445, 1989.
9. Ash SR: The allient dialysis system. Semin Dial 17: 164–166, 2004.
10. Ward RA: Ultrapure dialysate. Semin Dial 17: 489–497, 2004.
11. Meira FS, Poli de Figueiredo CE, Figueiredo AE: Influence of sodium profile in preventing complications during hemodialysis. Hemodial Int 11(suppl 3): S29–S32, 2007.
12. Rosenbaum BP, Ash SR, Wong RJ, et al: Prediction of hemodialysis sorbent cartridge urea nitrogen capacity and sodium release from in vitro tests. Hemodial Int 12: 244–253, 2008.
13. Rosenbaum BP, Ash SR, Carr DJ: Predicting dialysate sodium composition in sorbent dialysis using single point and multiple-dilution conductivity measurement. ASAIO J 51: 754–760, 2005.
14. Daugirdas JT: Simplified equations for monitoring Kt/V, PCRn, eKt/V and ePCRn. Adv Ren Replace Ther 2: 295–304, 1995.
Copyright © 2008 by the American Society for Artificial Internal Organs