Clearance of Selected Plasma Cytokines with Continuous Veno-Venous Hemodialysis Using Ultraflux EMiC2 versus Ultraflux AV1000S

Sepsis, severe sepsis, and septic shock encompass a spectrum of clinical syndromes caused by a dysregulated host response to infection [1,2,3]. Sepsis affects 18 million people worldwide each year, with mortality rates of 30-50% even in state-of-the-art intensive care units, and its incidence is increasing, mainly due to the emergence of antibiotic-resistant microorganisms and an aging population. Despite substantial advances in our understanding of the host response to infection and its dysregulation in sepsis, the development of targeted therapeutic approaches has been severely restricted by the considerable heterogeneity of sepsis patients [4,5,6] and the lack of accurate diagnostic methods to (1) prospectively identify those patients who would benefit from a specific therapy and (2) stratify patients into biologically homogenous subgroups, for example, with respect to plasma cytokine levels.

Acute renal failure, a frequent consequence of sepsis, has been identified as an independent risk factor for mortality in sepsis patients and has been shown to increase the complexity and cost of care [7]. Extracorporeal therapies are mainly applied to support or replace kidney function and to ameliorate uremic complications in sepsis, but there is evidence that these approaches might have effects beyond renal support by depleting inflammatory mediators, such as cytokines, chemokines, and complement factors from the circulation [8,9,10,11]. A significant portion of cytokines in the circulation, however, are bound to plasma proteins or associated with their target cells [12,] and are therefore not directly accessible to depletion by renal replacement therapies. Still, the depletion of unbound portions of these factors from the circulation may support the restoration of immune homeostasis by inducing re-equilibration of free and bound cytokine fractions.

Unlike drugs targeting single factors, extracorporeal therapies can influence a wide range of molecules, and the concept of blood purification has evolved towards the depletion of a broad spectrum of inflammatory mediators. High cutoff hemofilters, in particular, are characterized by an increased pore diameter of around 0.008-0.01 µm [13,14], and have been reported to exhibit more efficient depletion of inflammatory cytokines as compared to standard high-flux membranes [15,16,17,18]. In the majority of these studies, high cutoff filters were applied in convective settings, such as hemofiltration or hemodiafiltration, where enhanced cytokine depletion was reported to be associated with a reduction of albumin, in particular at increased effluent rates [19]. Here, we compared the removal of selected cytokines under predominantly diffusive conditions using continuous veno-venous hemodialysis with high cutoff hemofilters (CVVHD-HCO) versus standard hemofilters (CVVHD-STD), both in vitro and in sepsis patients with acute renal failure. Our primary hypothesis was that the two treatment modalities would mainly differ with respect to the depletion of cytokines in a molecular weight range of 20-25 kDa, such as IL-6.

Fresh frozen human plasma was obtained from the Red Cross, Linz, Austria. Recombinant human interleukin (IL)-6, IL-8, IL-10, and tumor necrosis factor-alpha (TNF-α) were purchased from R&D Systems, Minneapolis, MN, USA. Unfractionated heparin was obtained from Gilvasan Pharma, Vienna, Austria. Dialysis fluid (Fresenius Medical Care, Bad Homburg, Germany) contained 33 mM bicarbonate, 138 mM Na⁺, 2 mM K⁺, and 0.5 mM Mg⁺⁺.

The high cutoff filter Ultraflux EMiC2 and the standard filter Ultraflux AV1000S (Fresenius Medical Care) were used in this study. Both are polysulfone filters with a surface area of 1.8 m² and a molecular weight cutoff of approximately 30 kDa for Ultraflux AV1000S and approximately 45 kDa for Ultraflux EMiC2, respectively.

Fresh frozen human plasma (2 L) was thawed, filtered to remove cryo-precipitates, supplemented with 5 IU/mL of unfractionated heparin, and spiked with recombinant human IL-6 (target concentration 1,000 pg/mL), IL-8 (500 pg/mL), IL-10 (300 pg/mL), and TNF-α (800 pg/mL). The cytokine clearance for both filters was determined at a flow rate of 300 mL/min in the blood and dialysate circuits (Fig. 1a). Samples were drawn from the blood circuit at the filter inlet (S_in) and outlet (S_out) as well as from the dialysate circuit at the filter outlet (S_f) after 0, 15, 30, 60, 120, 180, and 240 min, and stored at -80°C until further analysis. Three independent sets of experiments were performed. For each set of experiment, a separate batch of plasma was spiked with recombinant human cytokines, and new filter sets were used.

The sieving coefficient (SC) was determined using the experimental set up shown in Figure 1b with flow rates of 300 mL/min in the blood circuit and a filtration rate of 30 mL/min. Samples were drawn from the blood circuit at the filter inlet (S_in) and outlet (S_out) after 0, 15, 30, 45, and 60 min as well as from the filtrate (S_f). All samples were stored at -80°C until further analysis. IL-6, IL-8, IL-10, and TNF-α were quantified with enzyme-linked immunosorbent assays (R&D Systems, Minneapolis, MN, USA). The clearance (K) was calculated as

K = (V/t)·ln (C₀/C_t)

with V denoting the plasma volume, t denoting the time of sampling, and C₀ and C_t denoting the plasma concentration of a given cytokine at 0 min and at the respective sampling time. The SC was determined as the ratio of cytokine concentrations (C) in the filtrate and plasma, and calculated as

SC = 2·C_f/(C_in + C_out)

To assess the cytokine depletion in vivo, we performed a single-center, randomized, controlled clinical study comparing continuous veno-venous hemodialysis using EMiC2 (CVVHD-HCO) versus AV1000S (CVVHD-STD). The study protocol was in accordance with the principles of the Declaration of Helsinki. Informed consent and approval by the Ethics Committee of the University Hospital St. Pölten, Austria (GS4-EK-3/082-2012) were obtained from all donors before the onset of the study. In total, 30 patients suffering from sepsis with acute renal failure were randomized to either CVVHD-HCO or CVVHD-STD and treated for 48 h using a blood flow (Qb) of 200 mL/min. Patients aged below 18 years or pregnant patients were excluded from the study.

Anticoagulation was performed with citrate according to the Ci-Ca protocol (Fresenius Medical Care). Simplified Acute Physiology Score (SAPS III) and Therapeutic Intervention Scoring System (TISS) score, leukocyte counts, comorbidities, as well as the sites of infection were recorded as baseline characteristics. Blood samples were drawn pre- and post-filter at baseline, that is, at the start of CVVHD, and at 1, 24, and 48 h. All samples were immediately centrifuged at 2,000 g (15 min, 4°C), and the resulting plasma samples were stored at -80°C until quantification of IL-6, IL-8, IL-10, TNF-α, and albumin. In addition, samples were drawn from the effluent at the indicated time points and analyzed as described above to trace cytokine transfer across the membrane. Cytokines were quantified with a bead array using the Bio-Plex 200 system (Bio-Rad, Vienna, Austria). Albumin was quantified using the colorimetric bromocresol green albumin assay (Sigma-Aldrich, St. Louis, MO, USA). The hybcell antibody microarray (CubeDX, St. Valentin, Austria) was used to quantify C-reactive protein (CRP) and procalcitonin (PCT). Plasma was mixed with conjugate solution at a ratio of 1:1, and 100 µL of the prepared mixture was loaded into the hybcell and analyzed, according to the manufacturer's instructions.

Data were analyzed using the SPSS version 22 (SPSS Inc., Chicago, IL, USA). The unpaired, non-parametric Wilcoxon rank sum test was applied to compare groups (i.e., high cutoff vs. standard filters at a given time point). The paired, non-parametric Wilcoxon signed-rank test was used to analyze changes in CRP and PCT levels over time in clinical samples for a given filter. p values of ≤0.05 were considered as statistically significant.

The in vitro clearance of IL-6 (M_r = 26 kDa) and IL-10 (M_r = 17 kDa) was significantly higher for CVVHD-HCO as compared to CVVHD-STD (4.7 ± 0.2 vs. 0.8 ± 0.4 mL/min for IL-6; 5.8 ± 0.6 vs. 2.9 ± 0.8 mL/min for IL-10) at 240 min. The low molecular weight cytokine IL-8 (M_r = 8 kDa) was cleared equally well by both filters with 16.9 ± 3.1 mL/min for CVVHD-HCO and 17.0 ± 3.0 mL/min for CVVHD-STD, while the clearance of TNF-α (M_r = 51 kDa) was 2.4 ± 0.5 vs. 0.6 ± 0.3 mL/min for CVVHD-HCO and CVVHD-STD, respectively. In vitro SCs were significantly higher for EMiC2 as compared to AV1000S for all tested cytokines, as summarized in Table 1. The decrease of cytokine concentrations in the plasma pool over time is shown in Figure 2.

We performed a single-center, randomized, controlled study in patients with sepsis and acute renal failure to assess the depletion of IL-6, IL-8, IL-10, and TNF-α using two treatment modalities. Of the 30 patients included in the study, 14 patients were treated with CVVHD-HCO, while 16 patients received CVVHD-STD. Patient baseline characteristics are summarized in Table 2, and mean baseline cytokine concentrations as well as the range for individual cytokines are displayed in Table 3. Baseline levels of IL-6, IL-8, IL-10, and TNF-α were elevated in the majority of patients and ranged widely within the study population. To assess the transfer of IL-6, IL-8, IL-10, and TNF-α across the EMiC2 and AV1000S membranes during CVVHD-HCO and CVVHD-STD, the concentration of these cytokines was determined in the effluent and is displayed in Table 4 as percentage of the respective plasma cytokine concentrations. In agreement with the in vitro data, EMiC2 depleted IL-6 more efficiently than AV1000S, as patients treated with EMiC2 showed a significantly higher percentage of IL-6 in the effluent relative to plasma concentrations at 1, 24, and 48 h. High amounts of the low molecular weight cytokine IL-8 were detected in the effluent for both filters, while the concentrations of IL-10 and TNF-α in the effluent remained below 10 and 1% of the respective plasma concentrations. Mean plasma levels of all cytokines tested decreased over time in both groups as summarized in online supplementary Table 1 (see www.karger.com/doi/10.1159/000478965).

CVVHD-HCO and CVVHD-STD did not differ significantly with respect to albumin removal. The albumin concentration in the effluent relative to the plasma concentration was 1.8 ± 0.3% for EMiC2 and 1.4 ± 0.2% for AV1000S, respectively. CRP levels showed no significant alterations over the treatment period for both filters, while PCT decreased significantly over time for both treatment modalities (Table 5).

The application of high cutoff hemofilters encompass clinical conditions requiring an effective removal of substances in the molecular weight range of 20-50 kDa, such as middle molecules, myoglobin, free light chain immunoglobulins, or cytokines. Enhanced cytokine depletion with high cutoff filters has been reported in a number of studies in vitro and in vivo [20,21,22]. It has been proposed that the concurrent removal of multiple mediators from the circulation can contribute to restoring immune homeostasis by eliminating the peaks of cytokine blood concentrations in the early phase of sepsis (peak concentration hypothesis) [23], by effecting the depletion of cytokines due to re-equilibration of blood and tissue cytokine levels (threshold immunomodulation hypothesis) [24], by inducing increased lymphatic flow and a displacement of mediators to the blood compartment (mediator delivery hypothesis) [25], or by directly acting on inflammatory cells, such as monocytes or neutrophils (cytokinetic model) [26].

In patients with sepsis-induced acute kidney injury, high cutoff hemofiltration was found to be associated with improvement of physiological parameters, reduction in vasopressor requirements [16], as well as restored monocyte proliferation capacity [15]. Due to their increased porosity, however, the application of high cutoff filters may be associated with albumin loss, in particular, under convective conditions and at increased effluent rates [27]. We, therefore, used CVVHD to assess the depletion of IL-6, IL-8, IL-10, and TNF-α, covering a molecular weight range from 8 to 51 kDa, by high cutoff versus standard hemofilters under predominantly diffusive conditions. SCs and cytokine clearances were initially determined in vitro in spiked human plasma to circumvent intra-experimental cytokine generation, showing that the high cutoff filter was significantly more efficient in clearing IL-6 and IL-10, which are in the middle molecular weight range, whereas the low molecular weight cytokine IL-8 was cleared equally well, and the removal of TNF-α was negligible for both filters.

To confirm these findings in vivo, we performed a single-center, randomized, controlled clinical study involving 30 sepsis patients with acute renal failure. Baseline plasma cytokine concentrations ranged widely across the study population and were in the upper range of those reported for severe sepsis, substantiating previous data and confirming the heterogeneity of sepsis patients [17,28,29,30,31].

Only insignificant amounts of albumin were detected in the effluent for both filters, demonstrating the safety of the high cutoff filter with regard to albumin loss. The quantification of cytokines in the effluent confirmed the transfer of IL-6 and of IL-8 across both filters, with significantly higher IL-6 and IL-8 effluent concentrations for the high cutoff filter. Cytokine transfer remained stable over the treatment period of 48 h without being substantially affected by fouling or secondary membrane formation due to plasma protein deposition, and there was no indication of coagulation with either of the filters. While these results confirm the in vitro data with respect to the depletion of IL-6, the transfer of IL-8 was less efficient as compared to the in vitro experiments for the standard filter. We assume that this reduced transfer in vivo may be due to the association of IL-8 with plasma proteins or blood cells. Likewise, only minor amounts of IL-10 were detected in the effluent for both filters. IL-10 has been shown to be associated with heparan sulfate shed from the activated endothelium in sepsis, which may decrease its transfer across the filter in vivo. Overall, plasma levels of all tested cytokines decreased over the treatment period of 48 h with both treatment modalities. The significantly enhanced removal of IL-6 and IL-8 with the high cutoff filter as compared to the standard filter did, however, not result in significantly lower IL-6 and IL-8 plasma concentrations at 48 h. This is in agreement with recent data from a retrospective study using CVVHD-HCO in 28 patients with septic shock and acute kidney injury, where the average IL-6 serum concentrations across the study population were even found to increase despite enhanced IL-6 clearance, apparently due to an intradialytic release of IL-6 in some patients, exceeding its elimination by dialysis [17].

In conclusion, CVVHD-HCO exhibited a significantly higher clearance for IL-6 and IL-10 as compared to CVVHD-STD in vitro. We could, however, not confirm a differential effect on plasma cytokine levels for CVVHD-HCO after 48 h in vivo, highlighting differences between cytokine depletion in vitro using human plasma spiked with recombinant cytokines and the in vivo situation. This may be due to de novo cytokine release depending on the status of the patient, as well as to the association of cytokines with blood cells, plasma proteins, or circulating glycosaminoglycans in vivo, resulting in their reduced transfer to the effluent.

We are grateful to Claudia Schildböck, Ute Fichtinger, and Birgit Fendl for excellent technical support and to Michael B. Fischer for valuable discussions. This work was funded by the Christian Doppler Society (Christian Doppler Laboratory for Innovative Therapy Approaches in Sepsis).

The authors have no conflict of interest to declare.