A randomized prospective cross over study on the effects of medium cut-off membranes on T cellular and serologic immune phenotypes in hemodialysis

Abstract

Extended cut-off filtration by medium cut-off membranes (MCO) has been shown to be safe in maintenance hemodialysis (HD). The notion of using them for the control of chronic low-grade inflammation and positively influencing cellular immune aberrations seems tempting. We conducted an open label, multicenter, randomized, 90 day 2-phase cross over clinical trial (MCO- vs. high flux-HD). 46 patients underwent randomization of which 34 completed the study. Dialysate- or pre- and post-dialysis serum inflammatory mediators were assayed for each study visit. Ex vivo T cell activation was assessed from cryopreserved leucocytes by flow cytometry. Linear mixed models were used to compare treatment modalities, with difference in pre-dialysis serum MCP-1 levels after 3 months as the predefined primary endpoint. Filtration/dialysate concentrations of most mediators, including MCP-1 (mean ± SD: 10.5 ± 5.9 vs. 5.1 ± 3.8 pg/ml, P < 0.001) were significantly increased during MCO- versus high flux-HD. However, except for the largest mediator studied, i.e., YKL-40, this did not confer any advantages for single session elimination kinetics (post-HD mean ± SD: 360 ± 334 vs. 564 ± 422 pg/ml, P < 0.001). No sustained reduction of any of the studied mediators was found neither. Still, the long-term reduction of CD69+ (P = 0.01) and PD1+ (P = 0.02) activated CD4+ T cells was striking. Thus, MCO-HD does not induce reduction of a broad range of inflammatory mediators studied here. Long-term reduction over a 3-month period was not possible. Increased single session filtration, as evidenced by increased dialysate concentrations of inflammatory mediators during MCO-HD, might eventually be compensated for by compartment redistribution or increased production during dialysis session. Nevertheless, lasting effects on the T-cell phenotype were seen, which deserves further investigation.

Figure 1

The flow chart shows patient recruitment and inclusion in two participating dialysis units [1: Klinikum rechts der Isar; 2: MVZ KfH Dachau]. Reasons for dropouts are displayed next to horizontal arrows (→). After all, a total of 34 patients completed the whole cross over phase.

Figure 2

(A) Reports mean, and 95%-CIs of dialysate MCP-1 concentrations stratified by treatment (MCO = 1) for all 4 time-points (T1 = Baseline; T2 = after 90 days of treatment with MCO or high flux; T3 = after another 30 days of wash out; T4 = after another days of treatment of high flux or MCO, respectively). (B) Δ-values of serum MCP-1 were calculated as: “pre-(0 h)-serum level” minus “post-(4 h)-serum level”. (C, D) Pre- and post- hemodialysis MCP-1 serum levels stratified by treatment (MCO = 1) are displayed. Herein, Pre-HD MCP-1 is equivalent to long-term (3 months) effects of MCO versus standard high flux membranes. (A–D) Statistical comparison was done using a linear mixed effect model with subject ID (nested in sequence) as random effect. Treatment (MCO = 1), period and sequence were tested as main effects. A P value of < 0.05 was considered statistically significant.

Figure 3

Reports means and 95% confidence intervals. (A) Reports mean dialysate concentrations of inflammatory mediators at T1 (baseline) and T3 (after switching treatments) in the intention to treat population. Comparison of treatments: MCO versus high flux dialysis was done using unpaired ANOVA on n = 42 MCO treated versus 39 high flux treated patients × sample pairs, comparable results were obtained on the per protocol population using linear mixed models (not shown). (B) Mean YKL-40 serum levels were assayed at all 4 study timepoints in the ITT population after the dialysis session (4 h—sample). (C) Mean YKL-40 serum levels were assayed at all 4 study timepoints in the ITT population before the dialysis session (0 h—sample) (B, C) The thin connecting horizontal lines between timepoints (T1–T4) represent repeated measures from subject IDs within sequence 1 (MCO → wash out → high flux). A linear mixed effects model was built with subject ID, nested in sequence, as a random effect and treatment (MCO = 1), period and sequence as the main effects. A P value for treatment < 0.05 in the absence of carry over effect was considered statistically significant.

Figure 4

(A–C) Report mean and 95% CI intervals for leukocytes, lymphocytes and CD3+ cells as reported above stratified by timepoint and treatment (MCO vs. high flux dialysis). P values were obtained from linear mixed effect models with subject ID (nested in sequence) with treatment (MCO = 1), period and sequence as main effects. A P values of < 0.05 was considered statistically significant. For CD3+ cells as well as (B) and (C) flow cytometric analysis was performed on re-thawed cryopreserved PBMCs which had been collected before the dialysis session (0 h) at all four study visits. Cells were then either left untreated (B) or stimulated with PMA + Ionomycin (C)—viability was generally > 85%. Means and SD can be obtained from Table 4. For details with regards to the gating strategy see supplement.

Figure 5

Cross over design, timepoints and sample acquisition.