Mechanisms of solute clearance and ultrafiltration in peritoneal dialysis

INTRODUCTION — Chronic peritoneal dialysis (as with continuous ambulatory peritoneal dialysis [CAPD]) can be complicated by problems with either solute clearance (which occurs by solute diffusion from the plasma into dialysate) or ultrafiltration (which is driven by the osmotic gradient between the hyperosmotic dialysate and the plasma) [1]. The mechanisms of solute clearance and ultrafiltration across the peritoneal membrane will be reviewed in this topic review. The clinical issues that can arise are discussed separately. (See "Inadequate solute clearance in peritoneal dialysis" and "Management of hypervolemia in patients on peritoneal dialysis".)

GENERAL PRINCIPLES OF TRANSPORT ACROSS THE PERITONEAL MEMBRANE — The average surface area of the peritoneal membrane is between 1 and 1.3 m² in adults [2,3]. During peritoneal dialysis, it is principally the parietal peritoneum that participates in peritoneal transport since only approximately one-third of the visceral peritoneum is in contact with the dialysis solution at a given time [4]. In addition to the capillary surface area, the diffusion length between the dialysate and the mesothelium also plays an important role in the overall transport characteristics of the peritoneum.

There are three barriers between the dialysate in the peritoneum and capillary blood: the capillary wall, which is most important; the interstitium; and the mesothelial cell layer. The mesothelial cell layer does not constitute a major barrier to solute or water transport across the peritoneum, while the interstitium offers some resistance to solute transport that is mainly restricted to large solutes [5].

Pores for solute transport — According to the three-pore model of solute transport, the capillary wall consists of a system of pores of three sizes, which are size selective in restricting solute transport [6,7]:

●There is an abundance of small pores (average radius 40 to 50 Å) that mediate the transport of lower-molecular-weight solutes. The transport of these solutes is limited by the number of small pores.

●The large pores constitute <0.1 percent of the total number of pores but are much larger than the small pores (average radius >150 Å). The radii of the smaller pores are constant, while the larger pores vary in size.

●Ultrasmall pores (3 to 5 Å), which constitute the third pore, are discussed in the next section.

Aquaporin-1 and water transport — The peritoneal membrane contains ultrasmall-pore water channels (3 to 5 Å), called aquaporin-1, which constitute the third pore in three-pore models of solute transport [6,8-11]. This is the same water channel present in red blood cells and the proximal tubule in the kidney [12] but different from aquaporin-2, which is the antidiuretic hormone-sensitive water channel in the collecting tubule [13]. Aquaporin-1, which is permeable only to water, is present in the endothelial cells of the peritoneal microvasculature as the major water channel [9,10]. Peritoneal tissue also contains small numbers of aquaporin-3 and aquaporin-4 [14].

The aquaporin system is responsible for transcellular water transport induced by the osmotic gradient created by adding hypertonic dialysate to the peritoneum. On average, it accounts for approximately 40 percent of total capillary ultrafiltration, with the remainder occurring via the paracellular route between the cells [6,15]. As an example, cumulative ultrafiltration is reduced by approximately 50 to 60 percent in mice lacking the aquaporin-1 gene [16,17]. By comparison, the absence of aquaporin-4 had no effect on osmotic water transport. Unlike aquaporin-mediated transport, which is primarily determined by the osmotic gradient, the small-pore water transport is dependent upon nonosmotic determinants [18]. Aquaporin-1 may also be involved in vascular proliferation and leucocyte recruitment during inflammation [19]. Preliminary in vivo evidence from a rat model of peritoneal dialysis suggests that it is possible to modulate the water transport function of aquaporin-1 by a pharmacologic agonist, such as AqF026 [20].

Solute transport — Solute transport across the peritoneum occurs via diffusion down a favorable concentration gradient (eg, high blood urea concentration compared with none in the dialysate) or convection.

●Diffusion varies directly with the magnitude of the concentration gradient and inversely with the size of the solute.

●Convection or solvent drag refers to solute transport that occurs with ultrafiltration (ie, water transport induced by an osmotic gradient), a process that is mediated by the frictional forces between water and solute. For any given solute, convection is measured by the sieving coefficient (S). S represents the ratio between the concentration of the solute in the ultrafiltrate and its concentration in the plasma, assuming that net diffusion is 0. On the other hand, the osmotic effectiveness of a solute against a membrane is measured by the reflection coefficient. The reflection coefficient (rc) varies from 0, if the membrane offers no resistance to transport of the solute, to 1, for an ideal semipermeable membrane that is completely impermeable to the solute. The relationship between these two parameters for an isoporous membrane is given by the equation: S = (1 - rc).

Low-molecular-weight solutes — The transport of low-molecular-weight solutes (eg, urea and creatinine) across the peritoneum primarily occurs by diffusion. This process is size selective (small solutes diffusing at faster rates than the larger solutes), and the rate is dependent upon the concentration gradient, peritoneal surface area, and peritoneal permeability. Peritoneal transport, at least in theory, may also be affected by other factors, such as vascular supply and the peritoneal charge. However, their overall contribution to solute transport is minimal at best.

The size selectivity of the peritoneal membrane is often expressed by the term "restriction coefficient," in which a value of 1 implies absence of a size restriction barrier. Thus, the higher the value of the restriction coefficient, the lower the size-selective permeability of the peritoneum.

The concept of mass transfer area coefficient (MTAC) represents the theoretical maximal peritoneal clearance by diffusion at time 0 (ie, before any solute transport has occurred). At this time, the concentration gradient across the membrane is at its highest and then progressively falls as solute diffuses from a compartment of high concentration to one of low concentration. In one of the simpler equations, small-solute transfer is supposed to follow first-order kinetics, with the contribution of convection assumed to be negligible [21]:

where Vt is the drained dialysate volume; t is the time; D0 and Dt are dialysate solute concentrations at start and at time t; and P is the plasma solute concentration.

Modifications of this equation that correct for convection have been proposed [22,23]. The calculation of MTAC, which is a research tool, is usually performed after a four- to six-hour dwell since inaccurate values may be obtained during shorter dwell times due to a large convective component and other factors. The relationship between the more commonly used D/P ratios (after a four-hour dwell) (figure 1) and MTAC is generally linear. However, D/P values may overestimate or underestimate MTAC when the MTAC is very low or very high, respectively.

Transport across the peritoneal membrane also occurs in the opposite direction, from the peritoneal cavity to the vascular compartment, via a process that is mainly diffusive and, therefore, size selective. This is particularly true of osmotic agents used in dialysis solutions such as glucose, glycerol, and amino acids. As an example, approximately 66 percent of the total instilled quantity of glucose is absorbed independent of the dialysate concentration used [24]. For solutes with a higher molecular weight, this transport occurs via a combination of convection (predominantly through lymphatics) and transport to the peritoneal interstitium, via transmesothelial route (see 'Fluid transport' below). The relative proportion of convection to diffusion is 0.1 for a small molecule like glucose compared with 10 for a large molecule like Hb when administered intraperitoneally [25,26].

There is appreciable variation in the rate of solute transport among individuals as some are slow transporters, and approximately 15 percent are rapid transporters. It is now well appreciated that peritoneal dialysis patients have different peritoneal membrane transport characteristics. These differences are best classified and determined by use of the peritoneal equilibration test (PET) (figure 2). Rapid transporters reach urea and creatinine equilibration more quickly, and the dialysate volume falls after two hours due to continuous reabsorption of fluid from the peritoneal cavity. (See "Peritoneal equilibration test".)

The rate of low-molecular-weight solutes also varies with dialysate instilled volume, with increased clearance being observed with increased volume [27]. By comparison, the clearance of beta-2 microglobulin, a middle-sized molecule, does not appear to change with changing instilled volume.

High-molecular-weight solutes — Peritoneal transport of larger molecules occurs at a much slower rate. Thus, creatinine (mol wt 113) is slower than urea (mol wt 56), inulin (mol wt 5200) is slower than creatinine, and larger proteins cross the peritoneum very slowly (figure 1). Proteins with higher molecular weights, such as albumin, transferrin, and immunoglobulin G (IgG), utilize the large pores described above for transport across the peritoneal membrane. The exact mode of this process is a subject of debate, but both size-selective diffusion and convection appear to contribute [28]. Regardless of the mechanism, this process is sufficiently slow that serum proteins are present in low concentration in the dialysate, and equilibration with the plasma does not occur at clinically used dwell times. As a result, the clearance of these molecules approximates their MTAC.

Both peritoneal permeability and the surface area of the peritoneal membrane influence the transfer of these larger solutes. The transport of these solutes out of the peritoneal cavity primarily occurs via the subdiaphragmatic lymphatics and, to a lesser degree, the peritoneal interstitium [25,29]. This process is independent of molecular size [29].

Electrolytes — The transport of electrolytes, such as sodium, potassium, and bicarbonate, across the peritoneal membrane may have important therapeutic implications. The dialysate concentration of sodium shows a decline in the initial phase of a hypertonic dwell, followed by a gradual increase [30,31]. The initial stage in this phenomenon, which is known as sodium sieving, results from selective transcellular water transport through ultrasmall, water-selective aquaporin-1 pores, lowering the dialysate sodium concentration by dilution [32]. Thus, short exchanges with hypertonic dialysate can lead to hypernatremia [33].

The gradual rise in the dialysate sodium concentration in the latter phase of the dwell is due to diffusion of sodium from the circulation. The MTAC for sodium is 4 mL/min when hypertonic glucose (3.86 percent) is used as a dialysate [34], while values of 9 mL/min have been reported for chloride [35]. Despite having a lower molecular weight than urea (23 versus 60), the MTAC of sodium is significantly lower. Thus, sodium behaves as a larger molecule, probably due to hydration-induced alterations in its configuration. By comparison, the MTACs for urea and creatinine are approximately 16 and 9.4 mL/min/1.73 m², respectively.

The average MTAC for the clearance of potassium by diffusion is between 12 and 16 mL/min in patients on continuous ambulatory peritoneal dialysis (CAPD) [34,35]. Potassium may also be released from the peritoneal lining, resulting in MTAC values as high as 24 mL/min during the first hour.

Because of favorable concentration gradients, there is a net mass transfer of calcium and magnesium from the dialysate to the vascular compartment by diffusion. The standard dialysate concentrations of calcium and magnesium are 1.75 and 0.75 mmol/L, which are higher than the free (ie, unbound) concentrations in the plasma (1.25 and 0.55 mmol/L, respectively). When higher glucose concentrations are used, this net transfer is balanced by convective transport due to ultrafiltration [36,37].

The MTAC for bicarbonate (mol wt 61) is approximately 9.5 mL/min [38]. Bicarbonate loss in peritoneal dialysis is dependent upon the plasma bicarbonate concentration (with lower values minimizing diffusion from plasma into the dialysate) and convective loss secondary to ultrafiltration. Standard dialysate contains lactate as a source of buffer at a concentration of 35 mmol/L. The use of standard dialysate results in an initial net diffusive and convective bicarbonate loss into the dialysate; this is counterbalanced by the movement of dialysate lactate into the blood as the subsequent metabolism of lactate generates bicarbonate [36].

The acid-base status of a patient on peritoneal dialysis is determined by the relationship between total metabolic acid production (a function of protein breakdown) and net loss or gain of base (bicarbonate loss versus lactate gain). An increase in the dose of dialysis (ie, increased number of exchanges) results in a net gain of lactate that can cause metabolic alkalosis. This phenomenon reflects maximum transfer of lactate during the initial part of the dwell, when the dialysate lactate concentration is at its highest.

Similar principles apply to the acid-base status of patients using bicarbonate-based peritoneal dialysis solutions. In this setting, bicarbonate diffusion out of the dialysate is partially counterbalanced by convective bicarbonate loss into the dialysate.

Fluid transport — Net transcapillary ultrafiltration (TCUF) is determined by the difference between the cumulative fluid transfer into the peritoneal cavity by ultrafiltration and the uptake of fluid out of the peritoneum, predominantly through peritoneal lymphatics. TCUF of water occurs via the small pores in the peritoneal membrane and the ultrasmall transcellular water channels (aquaporin-1). As mentioned above, the ultrasmall pores contribute as much as 40 percent to the total filtered volume [15,16].

The TCUF rate is determined by the net difference of osmotic and hydraulic pressures and the ultrafiltration coefficient of the peritoneal membrane, and is governed by Starling's law. Thus the TCUF rate is equal to:

which can be expressed as:

where UFC is peritoneal ultrafiltration coefficient, which is determined by the peritoneal surface area and its hydraulic permeability; Δπ is the oncotic or colloid osmotic pressure gradient; s is the reflection coefficient,; ΔO is the crystalloid osmotic pressure gradient (primarily determined by glucose); and cap and perit refer to values in the capillary and peritoneal fluid, respectively.

The intraperitoneal pressure (Pperit) is the major determinant of the hydraulic pressure gradient across the peritoneal membrane. The intraperitoneal pressure is directly dependent upon the instilled dialysate volume [39]. It is also influenced by posture, being approximately 2 to 8 mmHg when supine [40,41], but as high as 20 mmHg in the upright position (eg, walking) [40]. In one study, for example, a 10 mmHg elevation in intraperitoneal pressure reduced net ultrafiltration by 1.1 mL/min due both to reduced TCUF and increased lymphatic absorption [41].

The peritoneal capillary colloid osmotic pressure in peritoneal dialysis patients is approximately 21 mmHg [42]. Dialysate glucose is the major determinant of the crystalloid component of osmotic pressure. During CAPD, mean values for the osmotic reflection coefficient of glucose are reported as being quite low (approximately 0.02 to 0.05) [35,43-46]. This value is dependent upon the pore size in the peritoneal membrane. The reflection coefficient for glucose is 1 (ie, no transport) across the ultrasmall pores but approaches 0 (ie, unrestricted transport) across the large pores.

The small-solute component of the osmotic pressure gradient is maximal at the start of the dwell and dissipates with time as the dialysate glucose concentration falls due to both absorption into the systemic circulation and dilution by ultrafiltration; the net effect is that the maximum TCUF occurs at the start of the dwell [47]. The mean TCUF for a four-hour dwell with glucose varies with the glucose concentration, being 1 to 1.2 mL/min for 1.36 percent glucose [48,49] and 3.4 mL/min for 3.86 percent glucose [48].

The peritoneal membrane is relatively impermeable to large molecules, such as proteins and the glucose polymer icodextrin  (figure 1). When such solutes are present in the dialysate, there is a net fluid flow into the dialysate, a process termed colloid osmosis, which is similar to the effect of plasma proteins to keep fluid in the vascular space. Such solutions utilize the small-pore system for fluid transport. The pressure gradient exerted by these solutions is higher than the commonly used 1.5 percent glucose-based dialysis solutions (even though they are isosmolar) but lower than 4.25 percent glucose-based solutions.

The main advantage of using such colloid solutions is a slower loss of gradient dissipation compared with that due to the absorption of glucose. This property provides the rationale for the use of 7.5 percent icodextrin solutions for inducing TCUF. The absorption of icodextrin is only approximately 20 percent during an eight-hour exchange, allowing ultrafiltration to be maintained over at least 12 hours [42,49-51]. The mean ultrafiltration rate obtained by such colloid osmosis is between 1.4 to 2.3 mL/min [50,52]. In a randomized, controlled trial, isosmolar icodextrin (282 mosmol/kg) produced 3.5 times more ultrafiltration at eight hours and 5.5 times more at 12 hours than conventional 1.36 percent glucose dialysate (346 mosmol/kg) and was of equivalent efficacy to hypertonic 3.86 percent glucose (484 mosmol/kg) [50]. Icodextrin has been approved by the US Food and Drug Administration (FDA).

At all times and with all solutions, TCUF is counterbalanced by peritoneal reabsorption, the main routes of reabsorption being via the lymphatics and back-filtration by colloid osmosis. The lymphatic absorption rate varies directly with the intraperitoneal pressure, being higher with higher pressures [40].

CLINICAL CORRELATES — The preceding discussion on solute clearance and ultrafiltration assumes significance when considering the common clinical problem of ultrafiltration failure. The frequency of this problem increases with time [53]. (See "Inadequate solute clearance in peritoneal dialysis".)

Patients with ultrafiltration failure are typically evaluated with a peritoneal equilibration test (PET) (see "Peritoneal equilibration test"). There are four major causes of ultrafiltration failure:

●Large peritoneal vascular surface area, leading to significant glucose absorption

●Impaired aquaporin-1-mediated water transport

●Increased lymphatic absorption

●Peritoneal adhesions, leading to a small peritoneal vascular surface area

Large or small peritoneal vascular surface area — The most common type of ultrafiltration failure is associated with high dialysate-to-plasma ratios of small-molecular-weight solutes and is presumably due to an increase in peritoneal vascular surface area [53,54]. Although a higher water transport rate is achievable with a large surface area, the concomitant increase in absorption of glucose dissipates the osmotic gradient relatively early, resulting in loss of ultrafiltration (figure 2). There is no uniform agreement as to what causes high permeability [55]. Most believe it is due to larger surface area; however, there may be other causes that are not well defined, such as increased blood flow during an episode of acute peritonitis. This abnormality is called type 1 membrane failure, and these patients are called rapid transporters [53]. The rate of this rapid transport increases over time (figure 3). (See "Rapid transporters on maintenance peritoneal dialysis".)

Conversely, the presence of vasculopathy, such as due to deposition of advanced glycosylated end products in the peritoneal microcirculation, can adversely impact net ultrafiltration. This may be due to the reduction of hydrostatic pressure leading to decreased small pore fluid transport [56].

Type 2 failure, on the other hand, is associated with reductions in both solute transport and ultrafiltration due to a small peritoneal vascular surface area. This disorder is rare, often occurring in conjunction with sclerosing encapsulating peritonitis [57]. (See "Inadequate solute clearance in peritoneal dialysis".)

Impaired aquaporin-1-mediated water transport — Impairment in aquaporin-1 (ultrasmall-pore) expression and, more often, function can contribute to ultrafiltration failure. Constant exposure to glucose may eventually lead to peritoneal neoangiogenesis (similar to retinal neoangiogenesis in diabetics) [58], leading to both increased peritoneal vascular surface area and transcellular glycosylation of aquaporin-1. The prevalence of impaired aquaporin-1-mediated water transport as a cause of ultrafiltration failure is not known but is probably very low.

Assessment of aquaporin-1 function — As mentioned above, the dialysate concentration of sodium falls during the initial phase of a hypertonic dwell, followed by a gradual increase [30,31]. The initial fall in sodium concentration, which is known as sodium sieving, results from selective transcellular water transport through the water-selective aquaporin-1 pores, lowering the dialysate sodium concentration by dilution [32]. Thus, the extent of sodium sieving during a hypertonic exchange can provide an estimate of the overall function of these water channels and aquaporin-1 function (figure 4).

The net ultrafiltration obtained during the first two hours of an exchange with 4.25 percent glucose is compared with 1.5 percent glucose. This difference becomes smaller with impairment of aquaporin-1-mediated transcapillary ultrafiltration (TCUF) [58].

Several direct methods to assess free water transport have been described. These include quantification of free water transport from a single, standard peritoneal function test and by using a 4.25 percent mini PET (one hour) [59,60].

Ultrafiltration in patients with impaired aquaporin-1 function occurs via nonchannel-mediated pathways, which, as mentioned above, are responsible for approximately 60 percent of water transport across the peritoneal membrane [15,16]. Ultrafiltration can be enhanced in this setting by using icodextrin in long-dwell exchanges. Alternatively, 4.25 percent glucose exchanges with short dwell times may be helpful to some extent.

Increased lymphatic absorption — A decrease in net ultrafiltration due to enhanced lymphatic absorption (normally accounting for 10 to 20 percent of fluid reabsorption [61]) can occur in association with a large peritoneal vascular surface area. In such a setting, all measures that are utilized to enhance ultrafiltration should be instituted, including short cycle times, high glucose concentrations, and use of supine dialysis (net ultrafiltration in the supine position is approximately 16 percent higher due largely to a lower intraperitoneal pressure) [62]. Macromolecule, albumin, or hemoglobin absorption from the peritoneal cavity is increased in patients with increased lymphatic absorption [47]. Increased duration of peritoneal dialysis does not appear to enhance lymphatic absorption [61].

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Dialysis".)

SUMMARY AND RECOMMENDATIONS

●The peritoneal barrier is composed of three layers, including the peritoneal mesothelium, interstitium, and the capillary endothelium. According to the three-pore model of solute transport, the capillary endothelium contains three different-sized pores, which are size selective in restricting solute transport. (See 'General principles of transport across the peritoneal membrane' above.)

●Aquaporin-1 is the smallest sized pore and is responsible for almost 40 percent of free water transport across the peritoneal membrane. (See 'Aquaporin-1 and water transport' above.)

●The transport of different solutes across the peritoneal membrane is based on diffusion and convection. (See 'Solute transport' above.)

●The transcapillary ultrafiltration (TCUF) rate is determined by the net pressure gradient, as well as by peritoneal membrane characteristics, and is determined by Starling’s law. Both crystalloids and/or colloid-based dialysis solutions may be used to provide the required osmotic or oncotic gradients across the peritoneal membrane. (See 'Fluid transport' above.)

●The rate of solute transport varies between individuals, which may have therapeutic implications for dialysis adequacy as well as fluid overload. (See 'Clinical correlates' above.)

●Ultrafiltration failure may be a result of alterations in the vascular surface area (larger vascular surface area makes more pores available for transport), which leads to the rapid dissipation of glucose gradient across the membrane and fluid retention. Selective loss of aquaporin-1 function and other factors can also lead to ultrafiltration failure. The cause of ultrafiltration failure can often be determined by conducting special studies on the peritoneal membrane. (See 'Clinical correlates' above.)