Eberhard Ritz
Background
Peritoneal dialysis (PD) is used by over 100,000 end-stage renal disease (ESRD) patients worldwide, accounting for approximately 15% of the dialysis population (1). The relative value of PD is still uncertain and contested due to a lack of randomized clinical trials comparing dialysis modalities. However, clinical experience with PD suggests that it is a useful early therapy in ESRD and is a necessary third modality to complement hemodialysis and transplantation.
Along with peritonitis and inadequate removal of uremic toxins, ultrafiltration (UF) dysfunction is a major clinical limitation of PD. Over time, PD patients develop an increased transport of low-molecular weight solutes across the peritoneal membrane with a concomitant decrease in ultrafiltration (2). The capillary wall of peritoneal blood vessels plays a key role as a barrier to solute transport (3). Therefore, neovascularization of the peritoneal tissues is a main determinant of the functional changes in the peritoneal membrane. This increased vascular surface area leads to an increase in glucose transport from the peritoneal cavity with rapid loss of the osmotic gradient. The role of increased fluid absorption from the peritoneum by tissue and lymphatics (4), along with dysfunction of aquaporins in the peritoneal vessels (5), is less well defined.
A recent randomized controlled trial of adequacy of dialysis in Mexico (ADEMEX), reports no difference in all-cause mortality between PD patients dialyzed to a weekly total Kt/V of 1.7 compared with the recommended target of 2.0 (6). An increase in deaths due to uremia and congestive heart failure in the group with the lower Kt/V in the ADEMEX study suggests that adequacy should still be monitored clinically (7). Moreover, these results indicate that more attention needs to be paid to overall management of PD patients, particularly with respect to volume control and preservation of peritoneal membrane ultrafiltration capacity.
Definition and Clinical Approach
Earlier studies defined UF dysfunction clinically as an increasing requirement for high osmolarity dialysate (3 or more 3.86% glucose solution exchanges daily) to maintain an edema-free target weight (8). More recent studies have utilized objective criteria with a 4-h peritoneal equilibrium test (PET) using a high-concentration glucose dialysate solution where UF failure is defined as a net UF <400 ml/4 h (9,10). The standard PET using a 2.5% glucose solution and UF failure defined as <100 ml/4 h net ultrafiltration probably overestimates the occurrence of ultrafiltration dysfunction (10).
The recommended clinical approach to UF dysfunction has been outlined by a recent ISPD committee report (11). The first step is the recognition of UF dysfunction and the clinical manifestation, extravascular volume expansion. In the absence of any widely used ancillary measure of extravascular fluid volume, this is accomplished through careful clinical evaluation. The next step is to eliminate easily remediable problems. These include patient factors (e.g., dietary compliance), physician factors (inappropriate PD prescription), and mechanical factors (catheter malposition, extraperitoneal leakage). After evaluation of these easily remediable causes of volume expansion, peritoneal membrane dysfunction is, by exclusion, the diagnosis.
Epidemiology
In various studies, the prevalence of UF dysfunction leading to technique failure has been reported between 1.7% (12) and 13.7% (13). These data are confounded by the difficulty in accurately diagnosing the cause of technique failure. Using a clinical definition of UF dysfunction, prevalence rates have been reported between 15.3% (14) and 30.9% (8). Ho-dac-Pannekeet et al. (10) used the recommended 4-h high osmolarity PET and found the prevalence of UF dysfunction to be 23% in a group of 68 stable PD patients.
Changes in the peritoneal membrane function over time have been documented in several studies and recently reviewed (15). The majority of longitudinal studies of peritoneal membrane function find an increase in low-molecular weight solute transport with time on dialysis and a coincident decrease in ultrafiltration (2,16). The onset of UF dysfunction appears between 2 (17) and 4 (2) yr after the initiation of PD. The inverse correlation between UF and solute transport has been clearly demonstrated in these studies. Lo et al. (18) demonstrated a centripetal effect of membrane function over time. Among 55 PD patients followed for a mean of 22 mo, there was an increase in solute transport in patients with initial low transport peritoneal membranes and a decrease in solute transport in high transporters over time. One possible explanation is a selective censoring by technique failure or death of high-transport patients while there is increased transport in the low-transport groups over time (15). This may also explain to some extent the apparent stability of peritoneal membrane function over the first 2 to 4 yr of PD.
Pathophysiology
Ultrafiltration across the peritoneal membrane is the net effect of the osmotic force pulling fluid into the peritoneal cavity and hydrostatic forces pushing fluid into the interstitium and lymphatic reabsorption. The membrane across which the osmotic force acts has been described as a theoretical three-pore model (19). Ultrasmall pores correspond to aquaporin channels located on peritoneal venules and capillaries (5) and usually supply 40% of the net ultrafiltration (20). Small-sized pores are the major conduit for solute and fluid ultrafiltration. Large pores are few in number and allow for macromolecular transport. The major determinant of the prevalence of the small pores is the vascular surface area, which explains why increased vascularization of the peritoneal tissues will lead to rapid glucose transport, loss of osmotic gradient, and loss of ultrafiltration (3).
The peritoneal membrane is composed of an outer mesothelial cell layer and a basement membrane overlying a thin interstitial layer of connective tissue and blood vessels. In a recent large peritoneal biopsy study, Williams et al. (21) analyzed samples from 212 subjects including normal individuals as well as uremic and peritoneal dialysis patients. There was an increase in the thickness of the submesothelial tissue with uremia, which further increased with peritoneal dialysis. The vascular sclerosis, recognized in other biopsy studies (22), was correlated with time on peritoneal dialysis. The number of blood vessels in the submesothelial tissue did not correlate with time on dialysis. However, there was a significant increase in vascularization in patients whose biopsy was performed at catheter removal due to membrane failure (21).
Increased Vascular Surface Area
Increased vascularization of the peritoneal tissue is the primary mechanism of ultrafiltration failure. Between 50 and 75% of patients with UF dysfunction will have increased vascular area measured by increased small solute transport on a standard PET (8,10). In animal models of peritoneal dialysate exposure, an increase in submesothelial vascularization correlates directly with glucose absorption and inversely with net UF (23). Furthermore, reduction of peritoneal vasculature by anti-angiogenic therapy led to an improvement in net UF (23). Human biopsy studies confirm the association between submesothelial vascularization and UF dysfunction (21,22).
Aquaporin Dysfunction
Aquaporin dysfunction is a possible cause of UF failure in some patients. Monquil et al. (24) and colleagues studied six PD patients with severe UF dysfunction, normal small molecular transport rate, and low interstitial/lymphatic absorption rate. They suggested that aquaporin dysfunction was the remaining possible mechanism for UF dysfunction. Hyperosmolar PD solutions (3.86% glucose) did not improve ultrafiltration, and five of six patients required transfer to hemodialysis. However, the ability to accurately measure aquaporin function using a standard peritoneal function test with hyperosmolar solution (3.86% glucose) has been questioned (20). Also, normal aquaporin expression by immunohistochemistry of peritoneal tissue was found in a patient with presumed aquaporin dysfunction by clinical evaluation (25). Alteration of the function of the aquaporin channel could explain this finding, but more work needs to be carried out in this area.
Increased Interstitial/Lymphatic Absorption
Fluid reabsorption from the peritoneal cavity into tissues and lymphatics is a recognized mechanism of UF failure (4) and accounts for approximately 25% of the cases of UF dysfunction (8,26). The interstitium of the peritoneal tissue consists mainly of collagen fibers and glycosaminoglycans. Alterations in the constitution of the interstitium may lead to changes in hydrodynamic properties and macromolecular transport. Hydration will tend to expand the extracellular matrix and increase fluid conductance (27). Risk factors for increased tissue and lymphatic reabsorption have not been identified, but recently, Fussholler et al. (28) identified an association between increased fluid reabsorption and time on dialysis.
Etiology
The observation that UF dysfunction increases with time on PD (16) has led to the conclusion that elements of the PD fluid are bioincompatible and damage the peritoneal membrane (Figure 1). Several studies have documented progressive UF dysfunction in patients even in the absence of peritonitis (2). Animal models have confirmed the detrimental effect of peritoneal dialysate on the peritoneum (29). Putative bioincompatible components of standard PD fluid include glucose, glucose degradation products (GDP), lactate buffer, or acidic pH. The bioincompatible nature of low pH and buffer is not clear, but it appears to be less important than glucose or GDP (30).
Figure 1. Components of peritoneal membrane dysfunction, components of ultrafiltration dysfunction, and outcomes of ultrafiltration failure.
Davies et al. (31) have shown that increased exposure to hyperosmolar glucose dialysate predated by 2 yr the onset of UF dysfunction (31). In this case control study, patients with increasingly permeable peritoneal membranes also had an early loss of residual renal function, which probably required increased use of glucose to replace renal ultrafiltration. Selgas et al. (32) identified a similar effect with increased glucose exposure and diabetes, both associated with early UF dysfunction.
GDP are reactive carbonyl compounds produced in peritoneal dialysate during the process of heat sterilization. GDP, such as methylglyoxal, glyoxal, and 3-deoxyglucasone, have been shown to be more potent inducers of advanced glycation end-products (AGE) than glucose alone (33). GDP have been shown to alter mesothelial cell function and proliferation (34) and to induce vascular endothelial growth factor (VEGF) production in vitro (35). In animal models, conventional dialysate fluid induced vasodilation and capillary recruitment in the peritoneal tissues, whereas GDP-reduced dialysate did not induce these changes (30).
Peritonitis is a major cause of peritoneal membrane dysfunction. There is an acute effect of peritonitis that lasts for a month after the infectious event and clinically appears as decreased UF with increased solute transport. Ates et al. (36) have suggested a subtle longer-term UF dysfunction after a single episode of peritonitis. Others report that one episode of peritonitis has little long-term effect (2). However, multiple or severe infective episodes may have a greater long-term impact on the peritoneal membrane function (2).
The mechanism of long-term changes after peritonitis episodes may be due to an upregulation of inflammatory cytokines or profibrotic cytokines, such as TGF-β and fibroblast growth factor (FGF) (37). In an animal model of acute peritonitis, Combet et al. (38) demonstrated increased expression of endothelial nitric oxide synthase (NOS) in the peritoneal tissues. This was associated with increased vascularization, increased protein loss, and UF failure. Using adenovirus mediated gene transfer of inflammatory cytokines IL-1β and TNF-α, longer term effects on vascularization and fibrosis can be induced in the peritoneum after transient cytokine overexpression and IL-1β appears to be more potent in inducing these changes (39). Therefore, there are pathways from acute inflammation to longer-term fibrosis and angiogenesis in the peritoneum that may explain the association between peritonitis and UF dysfunction seen in several observational studies.
The uremic environment may be associated with altered peritoneal membrane structure and function. In rats made uremic with subtotal nephrectomy, significant changes in the peritoneal tissues occurred, including increased AGE deposition, increased vascularization with increased expression of NOS, VEGF, and FGF-2 (40). The effect of uremia on the peritoneal tissues has been confirmed in peritoneal biopsy studies that included uremic, nondialyzed patients, which demonstrated increased submesothelial thickening in these patients (21).
In PD patients, the association between peritoneal membrane transport properties and markers of systemic inflammation has been controversial. Wang et al. (41) did not find any association between membrane transport and numerous markers of inflammation assessed. In a second study, high C-reactive protein and declining residual renal function were both associated with increasing peritoneal membrane transport status from initiation to 1 yr on dialysis (42). In a retrospective study, we found that patients with a high transport peritoneal membrane had low serum albumin before the initiation of PD and suggested this was the result of systemic factors such as inflammation or overhydration (43).
Outcomes
Several factors, including low serum albumin, age, comorbid disease, and residual renal function, are known to be predictive of survival in PD patients. High transport status at the initiation of dialysis has also been shown to be associated with decreased UF function and poor outcome (44). Acquired UF dysfunction is associated with decreased survival in several studies. Wang et al. (45) measured the peritoneal membrane function in a cross-section of 46 established PD patients and found increased solute transport to be inversely associated with fluid removal and survival. In a study of 125 incident PD patients, Ates et al. (46) demonstrated the impact of total fluid removal and sodium removal (including peritoneal and renal) on survival. From a large longitudinal database, Davies et al. (47) compared survivors and nonsurvivors and demonstrated that nonsurvivors had a progressive increase in small solute transport with time on dialysis.
UF dysfunction is also associated with other poor prognostic indicators such as clinical volume expansion (48) and diastolic hypertension (49). Volume expansion in PD patients is associated with hypoalbuminemia (50), and adjusting dialysis prescription to increase UF can reduce volume expansion and increase serum albumin (51).
In summary, there is evidence that UF dysfunction is associated with decreased survival in PD patients. UF dysfunction is associated with volume expansion, and this may be associated, through hypertension, left ventricular hypertrophy, preexisting co-morbid disease, systemic inflammation, and hypoalbuminemia, with increased mortality and morbidity, primarily through cardiovascular outcomes (Figure 1).
Treatment
The treatment of UF dysfunction in PD patients involves three steps. First, the problem of UF dysfunction must be recognized through regular clinical volume assessment and routine measurement of peritoneal membrane function, including net UF (11). Prevention, with protection of the peritoneal membrane from injury due to peritonitis and bioincompatible dialysis solutions, is an important second step. Finally, treatment strategies for established UF dysfunction are limited at present, but they should expand with further research in this area.
Icodextrin
Icodextrin is a promising dialysate solution for improving UF function. This glucose polymer solution has been demonstrated to be beneficial in several clinical studies (Table 1). Icodextrin is a colloid and appears to promote UF through small pores in the peritoneal membrane and is therefore effective in PD patients with increased peritoneal vascular area or in the setting of acute peritonitis (52). In large randomized studies in which icodextrin was used in the nighttime dwell in CAPD patients (53) or the daytime dwell in APD patients (54), a similar UF rate was observed to that obtained with a hyperosmolar glucose solution (3.86% glucose). In a smaller observational study, the use of icodextrin in APD patients led to decreased use of hypotensive medication, fall in BP, and decreased extravascular fluid based on bioelectrical impedance analyses (55).
Table 1. Treatment trials using icodextrin and biocompatible solutions for ultrafiltration dysfunctiona
Two studies have assessed the value of icodextrin in patients with UF failure (Table 1). The first prospective study took 17 patients with impending technique failure due to UF dysfunction and prescribed icodextrin (56). Net daily UF increased by 570 ml, BP improved, and technique survival was prolonged by 1 yr. Patients who benefited the most from icodextrin were those who initially had very low UF.
Biocompatible Solutions
Biocompatible solutions are produced in two chambered bags so that glucose is separated from the electrolytes at a very low pH. This prevents the formation of GDP during heat sterilization (33) and allows for adjustment of final pH and buffer.
Conventional PD solutions have detrimental effects on cellular proliferation, function, and inflammation, which are reversed by more physiologic solutions (34,57). Animal studies using more physiologic solutions have demonstrated reduced AGE deposition and preserved UF (58).
Several clinical trials of physiologic solutions have been performed, and safety and tolerance have been demonstrated (Table 1) (59). In a 6-mo trial of 106 PD patients randomized to conventional or physiologic solution, Tranaeus (60) found an increase in UF (150 ml/d) with stable peritoneal membrane characteristics. In a voluntary 6-mo extension of this study, significantly lower peritonitis rates were reported in the physiologic solution group. A common finding among clinical trials of physiologic solutions is an increase in effluent CA125, suggesting improved mesothelial cell viability and a decrease in effluent hyaluronan concentration, which is interpreted as a decrease in peritoneal inflammation (61,62)
Other Alternate PD Solutions
Several other dialysate fluids have been tried in an attempt to increase UF. Amino acid solutions represent both an alternate osmotic agent and a source of nutrition. However, as an osmotic agent, amino acids appear to offer similar UF potential as conventional 1.5% glucose solutions (63). Glycerol has the advantage of avoidance of glucose; however, it is rapidly absorbed from the peritoneum. In hyperosmotic solutions, glycerol can provide equivalent UF to glucose (64). Low-sodium solutions have been used in an attempt to increase salt, and therefore water, removal. The results have been inconsistent (65).
Alteration of the Interstitial Compartment
Hyaluronan is a glycosaminoglycan that is present in the interstitium of the peritoneum. In animal models, hyaluronan has been shown to increase net ultrafiltration and decrease markers of inflammation (66,67). These animal studies have also shown a decrease in protein transport into the peritoneal fluid. Aside from demonstrated antiinflammatory properties, hyaluronan is felt to effect the interstitium and fluid absorption from the peritoneum (68).
Small observational studies in humans have been carried out to study the surface active phospholipid phosphatidylcholine and have demonstrated an increase net UF when delivered orally (69) or intraperitoneal (70).
Vasoactive Agents
Several agents with vasoactive properties have been studied. Nitroprusside is a nitric oxide donor and therefore increases the peritoneal vascular surface area and large molecule permeability. These effects would presumably have a negative impact on UF. However, when combined with icodextrin, nitroprusside increased net UF along with small and large molecule clearance in a small observational study in stable PD patients (71). On the other hand, inhibition of NOS should reduce peritoneal vascular surface area and permeability and therefore improve UF. This has been demonstrated in animal studies (72). Finally, angiotensin-converting enzyme inhibition is known to have effects on vasculature and should potentially be beneficial in preserving UF. This has been suggested in animal models (73), but not confirmed in small human studies (74).
Alternative Dialysis Strategies
Alteration of dialysis strategy is another potential technique to treat UF failure. Peritoneal rest has been advocated as an effective strategy and has been demonstrated in a clinical trial of 16 patients who were converted to HD for 4 mo (75). Net daily UF increased from 490 to 880 ml on re-initiation of PD. Increasing volume removal using furosemide or hemofiltration appeared to restore the peritoneal membrane UF capacity in a small study (76).
Automated PD (APD) has been suggested to be a useful therapy in patients with high-transport peritoneal membranes (77). The decreased dwell time with APD will maximize the UF using standard glucose solutions. However, observational studies suggest that APD patients have less sodium and fluid removal with resultant higher BP (78) and increased risk for left ventricular hypertrophy (79).
In summary, icodextrin has the potential to increase UF in patients with increased peritoneal vascular surface area or in the setting of peritonitis and may extend technique survival in patients with impending UF failure. Physiologic solutions hold great promise in the protection of peritoneal function and inflammatory response, but further clinical evidence is required. Pharmacologic agents have not been studied in any reasonably sized clinical trials; however, alteration of the interstitial compartment with modification of fluid reabsorption from the peritoneum may have significant benefits.
Conclusions
The ADEMEX study (6) and reanalysis of the CANUSA data (80) indicate that aspects of PD patient care other than solute removal need to be addressed. UF dysfunction is a major impediment to long-term PD therapy. Volume expansion, which progresses with changes in the peritoneal membrane transport properties and with loss of residual renal function, has a significant impact on the prevalence of cardiac disease in the PD population. UF dysfunction impacts both mortality and technique survival in PD patients.
Ongoing research has led to a further understanding of the mechanisms of UF failure, including increased vascular surface area and rapid glucose absorption, aquaporin dysfunction, and fluid absorption through interstitium and lymphatics. The etiologic agents involved (severe peritonitis, glucose, and bioincompatible dialysate) have been better defined. Future treatment strategies will include better detection of UF dysfunction and volume expansion, prevention of peritoneal membrane damage through new solutions, protection of RRF, and eventually, treatment of established UF failure using altered dialysis prescriptions and solutions, and perhaps pharmacologic intervention. Along with in vitro and animal model work to further define elements of UF failure, well-designed randomized trials are required to answer key clinical questions and improve outcomes for PD patients.
Acknowledgments
We would like to thank the many excellent researchers in peritoneal dialysis around the world whose ideas and insights have been adopted and form the basis for this review. P. Margetts is supported by the Canadian Institutes of Health Research.
- © 2002 American Society of Nephrology