Hemodialysis (HD) remains the most common renal replacement modality pretransplantation. Indications for HD are listed in Table 9.1 (4). Prior to transplantation, patients are generally ultrafiltered to reach their target weight and to optimize electrolytes and acid-base status. Limited data has emerged to help define the appropriate HD prescription to be delivered before surgery (see below). However, specific aspects of the hemodialysis prescription, particularly the biocompatibility of the dialyzer membrane, have been the focus of some research. Biocompatibility is a unifying term to encompass the complex interactions of the cellular and humoral components of blood with the artificial membrane and is dependent upon the chemical composition and permeability of the dialyzer membrane as well as the dialysate composition and temperature (5, 6, 7, 8, 9, 10, 11, 12, 13). Several biochemical consequences can occur in the setting of bioincompatibility (10) (Table 9.2), including abnormalities in leukocyte chemotaxis and oxidative metabolism, impaired expression of interleukin (IL)-2 receptors by peripheral blood mononuclear cells with cuprophane membranes, formation of platelet microaggregates and degranulation, and complement-mediated destruction of red blood cells (13,14). Concern over the chemical composition of membranes may exist during the
pre- and posttransplantation due to the purported relationship between inflammation and acute renal failure (10). The degree of serum complement activation based on dialyzer type is illustrated in Figure 9.1 (15). There are several categories of dialyzer membranes: (a) unsubstituted cellulose (e.g., cuprophane); (b) substituted cellulose (e.g., hemophane, cellulose di-/triacetate); and (c) synthetic (e.g., polysulfone, polyethersulfone, polyamide, polymethylmethacrylate, and polyacrylonitrile [e.g., AN69]) (Table 9.3) (10,16,17). Electron micrographs of several of these dialyzer materials are shown in Figure 9.2 (18). Other important features of dialyzer membranes include their capacity for clearance of inflammatory mediators of high molecular weight, such as cytokines and complement, as well as their ability to adsorb bacterial endotoxin. In this regard, C3a, C5a, C5b-9, IL-1, and tumor necrosis factor (TNF)-α have been advocated as serologic markers of biocompatibility (19). Within the past decade, more attention has been focused on the outcome of acute renal failure as it relates to membrane biocompatibility. While some researchers have reported improved renal recovery (20) and improved survival with synthetic versus cellulosic membranes (21), controversy still exists, as others (22, 23, 24) have found no such survival benefit (25). Subramanian and coworkers (25) constructed a meta-analysis of eight trials (N = 867), which revealed that the cumulative odds ratio for survival was 1:37 (95% confidence interval [CI]: 1.02 to 1.83, p = 0.03) in favor of synthetic dialyzer membranes. The survival advantage was limited to comparison between synthetic and unsubstituted cellulose membranes (i.e., cuprophane). In addition, a trend toward improved renal recovery was observed with synthetic membranes (odds ratio 1.23; 95% CI: 0.90 to 1.68, p = 0.18) (25). Van Biesen and coworkers (26) reported a trend toward greater recovery of renal allograft function, measured as reduction in serum creatinine to below 50% of pretransplantation level (T_{1/2 Scr}) in patients treated with synthetic membranes (26). Van Loo and coworkers (2) evaluated early allograft function in 44 HD patients in the 24 hours immediately preceding surgery and stratified them according to use of complement-activating (i.e., lower biocompatibility) cuprophane membranes versus synthetic (i.e., biocompatible) polysulfone membranes and according to the presence or absence of ultrafiltration. These 44 patients were also compared to 13 patients who had received HD more than 24 hours prior to transplantation. The authors observed that the T_{1/2 Scr} was 3.1 ± 2.9 days with polysulfone membranes versus 7.4 ± 7.9 days with cuprophane membranes (p <0.05) and 2.7 ± 2.0 days in those who did not receive ultrafiltration versus 7.1 ± 7.7 days for those who did (p <0.01). On the other hand, T_{1/2 Scr} tended to be lower (2.8 ± 1.7 days) in those receiving ultrafiltration more than 24 hours prior to transplant, regardless of membrane type. The authors concluded that early graft function was adversely affected by provision of HD with bioincompatible membranes and ultrafiltration within 24 hours of transplant and that, if possible, elective HD prior to renal transplantation should be avoided. However, provision of ultrafiltration alone may have a deleterious effect on early graft function, as patients dialyzed with polysulfone membranes had a T_{1/2 Scr} of 4.1 ± 3.5 days in the presence of ultrafiltration versus 1.7 ± 0.8 days in the absence of ultrafiltration (p = NS). In addition, the mean serum creatinine concentration was 3.0 ± 3.4 mg/dL with polysulfone membranes and ultrafiltration versus 1.1 ± 0.3 mg/dL with polysulfone membranes and no ultrafiltration (p <0.01). The authors postulated that ultrafiltration immediately preceding transplantation worsened effective circulating volume depletion and augmented renal hypoperfusion in a perioperative ischemic environment where autoregulation fails to maintain glomerular filtration. Moreover, these authors suggest a role for complement activation and free radical production on the outcome of renal allograft function (2). Woo and coworkers (3), on the other hand, did not find evidence to support an influence of membrane biocompatibility on outcomes following renal transplantation, an observation that has been supported by others (27,28). These investigators randomized 41 patients with posttransplant oliguria (defined as less than one liter of urine output per day) to receive intermittent hemodialysis with either cuprophane (N = 18) or polysulfone (N = 23) filters, which were all low flux and matched for urea clearance, surface area, and ultrafiltration characteristics. Following exclusion of five patients with primary graft nonfunction (three from the cuprophane group and two from the polysulfone group, p = NS), 36 patients were evaluated for the primary endpoint of date of last dialysis prior to establishment of diuresis (mean 10 [range 3 to 19] days in the cuprophane group versus mean 14 [range 1 to 26 days] in the polysulfone group, p = 0.3). Patients dialyzed with polysulfone membranes required a greater number of dialysis sessions (mean 7 [range 1 to 13]) versus those dialyzed with cuprophane membranes (mean 10 [range 3 to 19]) days (p = 0.03). In addition, there was a trend toward higher serum creatinine concentration at 1 month in the polysulfone group (mean 3.5 mg/dL) versus the cuprophane group (mean 1.9 mg/dL) (p = 0.1), and there was a trend toward a greater number of acute rejection episodes per 100 days of dialysis dependency in the polysulfone group (3.7 episodes) versus the cuprophane group (1.7 episodes) (p = 0.1) (3). The reasons for these findings are unclear; however, current practice in most institutions favors the use of synthetic, biocompatible membranes.

Peritoneal dialysis (PD) is provided to approximately 15% of ESRD patients. This renal replacement modality is favored over HD by some authors, because it provides hemodynamic stability, continuous clearance of uremic solutes, and greater independence to patients who rely on it (29). Most researchers have found favorable outcomes in the PD population in the setting of renal transplantation (26,30, 31, 32). For example, Bleyer and colleagues (30) compared transplantation outcome parameters such as urine production in the first 24 hours, need for dialysis within the first week after transplantation, and the incidence of acute allograft rejection in PD and HD patients. They used data from the United Network for Organ Sharing (UNOS) for 10,584 dialysis-dependent cadaveric graft recipients from April 1994 through December 1995. Compared with HD patients, PD patients were on dialysis for less time prior to transplantation, were more likely to be white, and had more favorable human leukocyte antigen (HLA) matching and lower panel reactive antibody (PRA). Cold and warm ischemia times were similar between the two groups. After adjustment for comorbid variables such as age, gender, race, HLA mismatch, time on dialysis, PRA, and cold and warm ischemia times, the percentage of patients who did not produce urine within the first 24 hours after surgery were 8.3% in the PD group and 11.9% in the HD group (p <0.001), respectively. Similarly, a greater percentage of HD patients required dialysis during the first week after transplant (28.6% in HD group vs. 20.0% in PD group, p <0.001). These findings could be explained by the existence of residual renal function in PD patients. However, the percentage of patients treated for acute rejection and those with nonfunctioning allografts at the time of hospital discharge were no different between HD and PD patients. The authors offer several possible reasons for these differences, including differences in cytokine production, shorter technique survival in PD (and, therefore, shorter time to transplantation), higher average hemoglobin levels in PD (reducing the likelihood of transfusion-related increases in PRA), and higher prevalence of white race in PD patients (30). Similarly, Vanholder and colleagues (32) found a lower incidence of delayed graft function (DGF) in a case control study of 234 patients (PD 27/117 = 23.1%; HD 59/117 = 50.4%, p <0.001) who were matched for age, sex, HLA compatibility, and cold ischemia time. Highly sensitized patients (PRA >85%) were excluded, but there was no difference in PRA status or warm ischemia time between HD and PD groups. While PD patients were, on average, treated with dialysis for a shorter time period prior to transplantation, the difference was not significant when compared with HD. Moreover, there was no difference in surgical or infectious complications posttransplantation between the two groups. Modality-specific complications were encountered more frequently in PD (peritonitis n = 7, tunnel/exit site infection n = 11, ascites n = 1) compared with HD (arteriovenous fistula occlusion n = 2) (32). Patient and allograft survival were no different in either group after 6 weeks or 6 months. Of note, PD patients developed acute rejection more frequently (80 of 117 PD patients vs. 61 of 117 HD patients, p <0.05) despite a similar number of patients receiving antithymocyte globulin in both groups (n = 43 each group). The authors did not conclude why this could have occurred, except to postulate that PD patients may be more immunocompetent than HD patients. Interestingly, the role of T-cell lymphocytes in ischemia-reperfusion injury of the kidney has recently been highlighted (33), and an immunodeficient state has been implicated
in both HD and PD patients (34). However, recent studies suggest that T-cell counts increase in PD relative to HD and that cytokine production by mononuclear cells is comparatively less in PD patients (35, 36, 37, 38, 39). Van Biesen et al (26) conducted a retrospective trial comparing posttransplant outcomes in 40 PD patients and 79 HD patients and found that the number of days needed to reach a serum creatinine concentration that was 50% below that before transplantation (T_1/2Scr) was more commonly greater than 5 days in the HD patients (22.7%) than PD patients (5%) (p = 0.01), which is important considering work by other investigators (40) to suggest that delayed function extending more than 6 days postoperative portends poor outcome. These researchers offered several possible reasons for this observation, including a stable, expanded volume status in PD patients; preservation of residual renal function in PD patients; difference in serum levels of vasoactive substances between PD and HD patients; and use of unmodified cellulose membranes in the HD patients studied (26). Fontan and coworkers (31) conducted a retrospective, single-center trial evaluating 56 PD patients and 58 HD patients (39 HD patients received grafts from the same donor as a PD patient, and 19 received a kidney from a different donor but were transplanted next to the one performed on the corresponding PD patient) and found favorable outcomes with the former modality, including greater likelihood of initial graft function (PD: 75.9% vs. HD: 50%, p <0.05) and shorter time to dialysis independence (PD: 7.8 ± 3.9 days vs. HD: 16.8 ± 8.0, p <0.025). Hospitalization length of stay, occurrence of acute rejection, and occurrence of infection within 1 month were not different between groups; however, patients who were receiving HD as a pretransplantation modality received more blood transfusions (9.2 ± 8.6 units vs. 5.8 ± 5.6 units, p <0.05) and had longer cold ischemia times (22.3 ± 7.5 vs. 19.4 ± 6.8 hours, p <0.05) than their counterparts who had received PD (31). An expanded retrospective analysis conducted at the same institution also found no difference in the incidence of primary allograft thrombosis between patients on PD and HD, despite prior evidence to the contrary (41, 42, 43, 44). Finally, Joseph and colleagues (42) also found a greater incidence of DGF in HD (58/117) vs. PD (56/183) (p = 0.01) but no difference in the occurrence of early (before 90 days) or late (after 90 days) acute rejection following transplantation (42).

The conventional HD procedure has been modified to remove anti-A and anti-B antibodies in preparation for ABO-incompatible renal transplantation (45). In cases of hyperacute rejection in accidental ABO-incompatible transplantation, plasma exchange followed by immunoadsorption has been used to extend the life of the allograft and to prevent need for explant (45, 46, 47, 48). As a result of this work, other authors have utilized plasma exchange and immunoadsorption in cases of ABO-incompatible living kidney donor transplantation and have demonstrated no difference in 5-year outcome when compared with ABO-compatible transplants (49,50). As an alternative to apheresis, Hout and colleagues (45) sought to modify the hollow fibers of traditional dialyzer filters by binding anti-A- and anti-B-specific antigen to the luminal surfaces and, thus, eliminate the additional step of separating plasma from whole blood, as in plasma exchange. Hemophane filters were exposed to a protein coupling solution (or bovine serum albumin as control), which was recirculated through the blood compartment at room temperature. Exposure of 100 mL of type O whole human blood resulted in a reduction of the anti-A antibody titer from 8 to 1 and the anti-B antibody titer from 16 to 1 after 30 minutes, while no antibody removal was observed with the control filters (Fig. 9.3). Further assays using anti-A- and anti-B-specific antigen from different blood donors suggested that anti-A and anti-B antibody titers in 300 to 400 mL of average to high titer type O blood could be reduced by 75% to 98% and that at least ten of these modified filters would be required to reduce titers of the entire blood volume of an average adult (45). The capacity of these filters may be increased in the future by more complete purification of the anti-A- and anti-B-specific antigens.