In vitro Validation of a Novel Disposable Remover to Remove Activated Leukocytes Generated During Cardiopulmonary Bypass: A Pilot Study

Introduction

Cardiopulmonary bypass (CPB) during cardiac surgery strongly increases risk of postoperative morbidity and mortality, primarily because it can induce systemic inflammatory response syndrome (SIRS),¹ which involves activation of leukocytes^2,3 that infiltrate into tissues,⁴ where they secrete excessive amounts of pro-inflammatory cytokines^5,6 that cause inflammatory reactions. In the past, glucocorticoids were widely used to dampen these inflammatory responses, yet they do not improve prognosis after on-pump cardiac surgery.^7,8 In another approach, the CytoSorb has been developed to remove pro-inflammatory factors from blood, but in several clinical trials, it failed to improve prognosis of patients who experience SIRS after CPB⁹ or of patients undergoing heart transplantation¹⁰ or surgery for acute infective endocarditis.¹¹

Another strategy is to remove from the blood the leukocytes that produce pro-inflammatory factors. Existing leukocyte filters have shown, at best, a transient ability to reduce leukocyte counts in blood after CPB without clinically significant benefits.^12–14 Our previous studies found that short-term–but not long-term–use of a leukocyte filter could attenuate SIRS and reduce infiltration of leukocytes into tissues,¹⁵ but that it could actually cause cellular membrane rupture and cytoplasmic leakage.¹⁶

Upon activation, inflammatory cells rapidly become adhesive by upregulating adhesion molecules on their surface in a polarized fashion, such that the molecules concentrate in an “adhesion domain” at the terminal end.^17,18 Our novel disposable leukocyte remover leverages this enhanced adhesive ability by incorporating RGD peptide (Arg-Gly-Asp) as a ligand on the membrane, which can stably bind to specific polypeptide groups within the adhesion domain at the terminal end of activated leukocytes,¹⁹ facilitating affinity adsorption. The resulting membrane (Chinese patent CN202310822538.X) is more adhesive and biocompatible than the existing one on which it is based. Furthermore, the membrane is connected at both ends to a hollow support structure, which prevents membrane contact during helical rolling inside the remover and maximizes the membrane surface available for blood contact. The membrane-support structure is integrated into a cylindrical disposable device (Chinese patent CN202211057321.6) that in turn can be integrated into a CPB circuit. Using an extracorporeal circuit, we assessed here the ability of the novel leukocyte remover to selectively remove activated leukocytes from the blood of patients who underwent bypass.

Methods

Extracorporeal Circuit and Patient Blood

A specially treated membrane with a surface area of 50 cm² was produced by the Institute of Blood Transfusion at the Chinese Academy of Medical Sciences in Chengdu, China, then rolled into a cylindrical shape, outfitted at both ends with guide tubes, and placed within a rigid housing (Chengdu Shangyuantai Biotechnology, Chengdu, China; Figure 1A). For the present study, the remover was integrated into an extracorporeal circuit containing a 3-L M/Y reservoir (Jingjing Medical Equipment, Beijing, China) and a Stockert S5 roller pump (Sorin, Mirandola, Italy; Figure 1B).

Figure 1 Design of the novel disposable leukocyte remover and its validation in an extracorporeal circuit in vitro. (A) Schematic of the remover, showing the outlet (1), cap (2), shell (3), membrane (4), inner support core (5), membrane support structure (6), and inlet (7). (B) The remover was integrated into a circuit. Blood from patients who had undergone surgery involving cardiopulmonary bypass was stored in a reservoir and circulated by a roller pump. The blood was sampled immediately before circulation and after 5, 10, 15 and 20 cycles. Activated cells were counted in samples using routine blood analysis and flow cytometry. Morphology of cells was analyzed in smears from blood prepared immediately before circulation and after 20 cycles. Adsorbed cells were analyzed on the membrane after 20 cycles with each of the seven blood samples. Abbreviation: PBMCs, peripheral blood mononuclear cells.

This is a substudy involving tissue samples obtained within the DIMOCS study, which was approved by the Biomedical Ethics Review Committee of West China Hospital, Sichuan University (No. 2022–262) and registered at Clinicaltrials.gov (NCT05400356). This study was conducted at the Cardiovascular Department of West China Hospital of Sichuan University between December 1, 2022 and March 1, 2023. Patients older than 18 years were included in this study and would not be included into the ongoing large clinical trial. The protocol followed by the guidelines of the Declaration of Helsinki. Written informed consent was obtained preoperatively from all patients and their relatives, and they were informed that they could leave the study at any time for any reason, without consequences.

From seven patients who underwent CPB, blood was harvested from the bypass circuit by flushing normal saline through the arterial, venous and suction tubes as well as the BB841 oxygenator (Medtronic, Minneapolis, USA) and the Type A perfuser (Wego Medical Equipment, Shandong, China). The resulting wash-out fluid (approximately 500 mL) was concentrated to 150 mL using an Xtra hemoconcentrator (Sorin, Mirandola, Italy) and added to the extracorporeal circuit. Blood from each patient was circulated for 10 min at a driving pressure of 100 mmHg and flow rate of approximately 300mL/min, each circulation cycle lasted approximately 30 seconds. The blood was sampled immediately before circulation and after 2.5, 5, 7.5, and 10 min in circulation, corresponding to roughly 5, 10, 15 and 20 cycles. Samples were subjected to routine blood testing on an automated analyzer (Sysmex, Kobe, Japan) as well as flow cytometry to detect the expression of activated cells and molecules (see below). After all seven samples had been circulated, the remover membrane was taken out and analyzed using scanning electron microscopy and immunofluorescence microscopy (see below).

Flow Cytometry of Activated Leukocytes and Molecules in Blood

Sampled blood was added to 8 mL of 0.85% NH₄Cl solution, allowed to sit at room temperature for 10 min, and centrifuged at 300 ×g at 4 °C for 10 min to lyse the erythrocytes. The supernatant was discarded, and the peripheral blood mononuclear cells (PBMCs) were resuspended in phosphate-buffered saline (PBS) containing 2% fetal bovine serum. The resuspension was centrifuged again at 300 g at 4 °C for 10 min, and PBMCs were resuspended in PBS to a concentration of 2 × 10⁶/mL. In separate tubes, equal numbers of PBMCs in aliquots of 100 μL were stained at room temperature in the dark for 30 min with an antibody against an appropriate surface marker to label granulocytes, monocytes, T lymphocytes, B lymphocytes and natural killer cells (Table 1). Samples were centrifuged at 300 g at 4 °C for 5 min, and the PBMCs were resuspended in 300 μL PBS, then analyzed on a FACSLyric flow cytometer (BD Biosciences, San Jose, CA, USA) using FlowJo 10.8.1 software (Treestar, Ashland, OR, USA) with manual gating and automated clustering methods. The manual gating was set to exclude debris, dead cells and doublets but retain single live cells for subsequent analysis (Figure 2A). In order to identify and visualize activated cell subpopulations, all live cells were clustered through an automated algorithm (Figure 2B). Subpopulations of leukocytes were plotted onto two-dimensional maps using t-distributed stochastic neighbor embedding (t-SNE)²⁰ and grouped into phenotypically homogeneous clusters using the “self-organizing map” routine (FlowSOM) in the automated algorithm.^21,22

Table 1 Antibodies for Flow Cytometric Identification of Activated Cells in Blood

Figure 2 Flow cytometry data analysis. (A) Manual gating in FlowJo software, which allowed cell selection based on FSC and SSC in a region of stable flow, while excluding debris, dead cells and doublets. Granulocytes, monocytes and lymphocytes were gated based on expression of CD45, then expression of molecules on the surface of each cell type was analyzed. (B) Dimensionality reduction and clustering analysis of flow cytometry data were performed using FlowJo software in conjunction with R in order to identify cell subpopulations whose counts decreased during circulation. Activated cell subsets were visualized in two dimensions using the t-SNE algorithm, with each point corresponding to a cell and their closeness indicating similarity. The t-SNE maps were analyzed using FlowSOM, which automatically clustered cells into phenotypically similar populations, which were color-coded to facilitate analysis of variations in cell abundance during circulation. Abbreviations: FSC, forward scatter; FCS files, flow cytometry standard files; NK, natural killer; SOM, self-organizing maps; SSC, side scatter; t-SNE, t-distributed stochastic neighbor embedding.

Blood Smears

Smears of blood sampled immediately before circulation and after 10 min of circulation were prepared and stained using a G1040 Wright-Giemsa staining kit (Solarbio, Beijing, China). After the smear dried, it was covered with 1 mL of solution A and allowed to sit for 1 min. Then 3 mL of solution B was added and mixed thoroughly with solution A, and the sample was allowed to sit for 3–10 min. The sample was rinsed with water and examined under an Axio Imager A2 LED/DL microscope (Carl Zeiss, Oberkochen, Germany).

Analysis of the Leukocyte Membrane and Adsorbed Cells

After all seven samples had been circulated, the membrane was taken out from the leukocyte remover and some of it was cut into square pieces of 0.5×0.5 cm², fixed in 2.5% glutaraldehyde for 1–2 h, rinsed 3 times for 5 min each in PBS containing 2% fetal bovine serum, and dehydrated by soaking sequentially through solutions of 30%, 50%, 75% and 90% anhydrous ethanol in ultrapure water for 5–10 min each time. Finally, the membrane was soaked three times for 5–10 min each time in pure anhydrous ethanol, dried in a CO₂ dryer, attached to a metal stage with conductive adhesive, sputter-coated with ion for 8 min, and examined under an Evo 10 scanning electron microscope (Carl Zeiss, Oberkochen, Germany) at magnifications of 2000–5000.

The remaining part of the membrane was cut into square pieces of 1.0×1.0 cm², fixed in 2.5% glutaraldehyde for 1–2 h, washed three times with PBS containing 2% fetal bovine serum, blocked for 30 min at room temperature with 5% fetal bovine serum, incubated with anti-CD11b antibody (diluted 1:3000) for 1 h at room temperature or overnight at 4 °C, washed in 2% PBS three times for 5 min each, incubated with secondary antibody for 1 h in the dark at room temperature, again washed in 2% PBS three times for 5 min each, and finally incubated with 4′-6-diamidino-2-phenylindole (DAPI) to stain nuclei. Samples were coverslipped using anti-fade mounting medium and examined under an A1rmp+ two-photon microscope (Nikon, Tokyo, Japan).

Statistical Analysis

Data were analyzed using SPSS 29.0 (IBM, Armonk, NY, USA). Normally distributed data were reported as mean ± standard deviation, and differences were assessed for significance using one-way ANOVA. Skewed data were reported as medians, and differences were assessed using the non-parametric K-samples test. In all cases, differences were considered significant if associated with P < 0.05.

Results

Blood for testing the disposable leukocyte remover was taken from seven patients who underwent valve replacement surgery or radiofrequency ablation involving CPB (Table 2).

Table 2 Clinicodemographic Characteristics of the Seven Patients Whose Blood Was Used in This Study

Changes in Blood Cell Counts

Before circulation, the blood after concentration contained (5.62±1.57)×10⁹/L leukocytes, (3.68±1.01)×10⁹/L neutrophils, (0.19±0.06)×10⁹/L monocytes, (100±25.23)×10¹²/L platelets, (75±11.00) g/L hemoglobin and (2.52±0.66)×10¹²/L red blood cells. The counts of each cell type were within normal reference ranges. Within 2.5 min of circulation, during which the blood passed approximately 5 times through the membrane, the leukocyte count dropped by 38%, after which it decreased by approximately 14% every 5 cycles. The levels of hemoglobin and the counts of platelets and red blood cells did not change significantly during 20 cycles (Table 3).

Table 3 Cell Counts in Blood as a Function of Number of Cycles in the Disposable Leukocyte Membrane Remover

Changes in Neutrophils

Flow cytometry showed that counts of neutrophils and of neutrophils expressing CD11b, CD54, CD64 and CD181 decreased by 36–39% during the first 2.5 min, after which they continued to decrease more slowly (Figure 3A). We identified three metaclusters of neutrophils whose counts decreased in t-SNE density plots during the 20 cycles (Figure 3B): metacluster 10 decreased significantly by 46% during the first 2.5 min, after which it declined more slowly; while metaclusters 14 and 17 significantly decreased, respectively, by 72% and 67% after 5 min and more slowly thereafter. All three metaclusters were neutrophil subsets expressing both CD11b and CD64 (Figure 3C).

Figure 3 Changes in counts of neutrophils. (A) Changes in the counts of total neutrophils and of subsets expressing the indicated surface molecules (n=7). (B) Changes in relative abundances of 20 clusters of neutrophils, defined based on t-SNE. (C) Metaclusters of neutrophils based on FlowSOM analysis. Rows correspond to metaclusters, while columns correspond to their expression of surface markers (blue, lowest expression; red, highest expression). Metaclusters whose t-SNE density decreased significantly during the 20 cycles are boxed in black. Box plots above and below on the right side of the panel depict metacluster counts as a function of cycle and metacluster position in the t-SNE plot. In the middle of the right side of the panel, peak expression plots of surface markers are shown for three metaclusters. C0, C5, C10, C15, and C20 indicate: before circulation and after 5, 10, 15 and 20 cycles, respectively. * P < 0.01, ** P < 0.05, ***P < 0.001, ****P < 0.0001 compared to C0. Abbreviations: Pop, populations; t-SNE,t-distributed stochastic neighbor embedding.

Changes in Monocytes

Flow cytometry showed that total monocyte count did not decrease significantly until 5 min, when it fell to approximately 33% of the pre-circulation value, after which it decreased by approximately 17% every five cycles. The total decrease affected primarily classical and nonclassical monocytes, because counts of intermediate monocytes did not change significantly during 20 cycles. Flow cytometry also showed that counts of monocytes expressing CD163, CD274 and CD284 decreased by 70–71% during the first 5 min, after which they declined more slowly (Figure 4A). We identified two metaclusters of classical monocytes whose counts decreased in t-SNE density plots during the 20 cycles (Figure 4B): metacluster 9 decreased significantly by 46% during the first 2.5 min, then more slowly thereafter; while metacluster 11 decreased during the 20 cycles, although the decrease did not achieve statistical significance. Both metaclusters were subsets of classical monocytes expressing both CD80 and HLA-DR (Figure 4C).

Figure 4 Changes in counts of monocytes. (A) Changes in the total number of monocytes and numbers of their surface activation molecules (n=7). (B) Changes in relative abundances of 20 clusters of monocytes based on t-distributed stochastic neighbor embedding (t-SNE). (C) Metaclusters of monocytes based on FlowSOM analysis. Rows correspond to metaclusters, while columns correspond to their expression of surface markers (blue, lowest expression; red, highest expression). Metaclusters whose t-SNE density decreased significantly during the 20 cycles are boxed in black. Box plots above on the right side of the panel depict metacluster counts as a function of cycle and metacluster position in the t-SNE plot. In the below of the right side of the panel, peak expression plots of surface markers are shown for two metaclusters. C0, C5, C10, C15, and C20 indicate: before circulation and after 5, 10, 15 and 20 cycles, respectively. * P < 0.01, ** P < 0.05 compared to C0. Abbreviations: Pop, populations; t-SNE,t-distributed stochastic neighbor embedding.

Changes in Lymphocytes

Counts of total T cells or of T cells expressing CD8 did not decrease significantly during the 20 cycles. In contrast, counts of T cells expressing CD4 decreased significantly by 35% after 5 min (Figure 5A). T lymphocyte t-SNE density maps indicated no reduction in cell subsets densities during circulation (Figure 5B). Counts of all natural killer (NK) cells did not decrease significantly during the 20 cycles, while counts of NK cells expressing CD57 decreased significantly by 39% during the first 2.5 min (Figure 5C). NK cell t-SNE density maps also showed no reduction in cell subsets densities during circulation (Figure 5D). Among B lymphocytes, counts of those expressing CD274 decreased significantly by 36% during the first 2.5 min, counts of those expressing CD38 decreased significantly by 40% in the first 5 min, and counts of those expressing CD80 decreased significantly by 41% during 20 cycles (Figure 5E). B lymphocyte t-SNE density maps revealed no decrease in cell subpopulation density during circulation (Figure 5F).

Figure 5 Changes in lymphocytes. (A) Changes in the total number of T cells and numbers of their surface activation molecules (n=7). (B) Changes in relative abundances of 20 clusters of T cells based on t-distributed stochastic neighbor embedding (t-SNE). (C) Changes in the total number of NK cells and numbers of their surface activation molecules (n=7). (D) Changes in relative abundances of 20 clusters of NK cells based on t-SNE. (E) Changes in the total number of B cells and numbers of their surface activation molecules (n=7). (F) Changes in relative abundances of 20 clusters of B cells based on t-SNE. C0, C5, C10, C15, and C20 indicate: before circulation and after 5, 10, 15 and 20 cycles, respectively. * P < 0.01, ** P < 0.05 compared to C0. Abbreviations: NK, natural killer; t-SNE, t-distributed stochastic neighbor embedding.

Effects of Remover on Leukocyte Morphology and Blood Cells

Blood samples before circulation contained neutrophils with ruptured cell membranes and prominent pseudopods, and such cells were nearly absent following circulation, indicating their efficient removal (Figure 6A). Circulation during 20 cycles did not appear to alter the morphology or membrane integrity of lymphocytes or monocytes, nor did it significantly affect levels of hemoglobin or counts of platelets or red blood cells (Figure 6B).

Figure 6 Effect of the remover membrane on morphology of blood cells and platelets. (A) Representative photomicrographs of different leukocyte types before circulation (left) and after 20 cycles (right). The red arrow marks a pseudopod. Magnification, 400×. (B) Levels of hemoglobin and counts of red blood cells and platelets during circulation (n=7). C0, C5, C10, C15, and C20 indicate: before circulation and after 5, 10, 15 and 20 cycles, respectively. Abbreviations: RBC, Red Blood Cell; PLT, platelet.

Analysis of the Remover Membrane

After 20 cycles with each of seven blood samples, the remover membrane appeared smooth and intact and contained numerous intact cells (Figure 7A), which immunostaining revealed to be activated leukocytes (Figure 7B).

Figure 7 Leukocytes adsorbed onto the remover membrane. The membrane was removed after 20 cycles with each of seven blood samples, then examined using (A) scanning electron microscopy; or (B) immunostaining against CD11b, a surface marker for activated leukocytes (green), as well as counterstaining of nuclei using DAPI (blue). Magnification factors are indicated at the far left. Scale bar, 100 μm. Abbreviation: DAPI, 4′-6-diamidino-2-phenylindole.

Discussion

Here we describe a novel leukocyte remover and validate it in vitro for its ability to remove, effectively and preferentially, leukocytes that have been activated during CPB. This study justifies further exploration of the novel leukocyte remover membrane as potentially superior to traditional membranes that have demonstrated limited effectiveness in enhancing prognosis following CPB.^23,24

Activated leukocytes are more adhesive than non-activated ones. Within the first few minutes of circulation, our remover demonstrated a significant ability to remove activated neutrophils, which are among the earliest cells in the bloodstream to participate in inflammatory responses. Levels of activated leukocyte molecules (CD11b, CD54, CD64 and CD181) on neutrophils, consistently declined in abundance with each cycle in circulation. These molecules serve primarily as adhesive and inflammatory chemotactic agents and thereby mediate inflammatory responses and associated tissue damage.^25,26 Elevated levels of CD64, notably, are a diagnostic indicator of SIRS.²⁷ In our research, we observed a significant reduction in counts of neutrophils expressing these markers during circulation, suggesting that the remover effectively targets activated neutrophils. Immunofluorescence staining revealed a significant accumulation of CD11b+ neutrophils on the membrane, corroborating its adsorption capabilities.

At the same time, the remover had negligible impact on counts of monocytes and lymphocytes during the first few minutes of circulation. Most monocytes and lymphocytes whose counts decreased during circulation were activated, such as monocytes expressing CD163 and NK cells expressing CD57. Expression of CD163, a receptor involved in inflammation and hemoglobin clearance, increases during inflammatory responses, making it a reliable marker of inflammation.²⁸ Lipopolysaccharide binds to a receptor comprising Toll-like receptor-4 (TLR4/CD284) and CD14 on the surface of monocytes to activate them.²⁹ Monocytes, as well as macrophages, express PD-L1 (CD274), a ligand for PD-1 (CD279), and levels of PD-L1 on monocytes are associated with risk of mortality in sepsis patients.³⁰ CD57 serves as a marker of NK cell maturity, potentially indicating terminal differentiation.³¹ Modest reduction of monocytes and lymphocytes during circulation, we propose that, the adsorptive removal of these cells during circulation may occur via the adhesion molecules CD11b and CD62L, which are expressed at low levels on the surface even in the absence of activation.³² Applying the leukocyte remover for shorter than 2.5 min may help minimize adverse impacts on immune cell counts. During this short period, our remover appears to target preferentially activated leukocytes, particularly neutrophils, while minimally affecting other leukocyte subsets, potentially making it superior to previously described membranes.¹⁶ In this way, the remover may help mitigate the postoperative rise in levels of inflammatory cytokines without compromising the body’s immune function.^33,34

Another reason to limit how long the remover is applied is to prevent the activation and lysis of leukocytes similar to that reported after prolonged application of a previously described leukocyte membrane.¹⁶ Such activation and lysis increase risk of inflammatory responses. In our experiments, the bulk of neutrophil removal occurred within the first 2.5 min, corresponding to approximately 5 cycles through the circulatory loop. Future work should optimize the cycling time to maximize leukocyte removal and improve safety simultaneously.

We did not observe substantial effects of the remover on counts or morphology of platelets or red blood cells, implying minimal interference with coagulation or oxygenation capacity of the blood. Furthermore, the blood cells that were not adsorbed showed similar morphology before and after passing over the membrane. Microscopic examination of the cellular morphology on the adsorptive membrane revealed that white blood cells remained intact, with no exposed nuclei detected, implying negligible leakage of proteases and pro-inflammatory substances.

We attribute the greater efficacy and selectivity of our membrane to the fact that blood flows over it rather than through it as in previously described membranes. Our membrane leverages the strong adhesiveness of activated leukocytes. Additionally, the hollow support structure of the membrane ensures that blood makes contact with both sides of the membrane as it flows through the remover, maximizing adsorption efficiency.

Our findings should be interpreted with caution given that concentrated residual blood may not behave identically to normal human blood and may differ significantly in concentrations of proteins and coagulation factors. This difference might affect the equivalence of concentrated blood to whole blood removal. Another limitation of our study is that we used a membrane with a surface area of only 50 cm², raising the question of whether the disposable remover will function similarly when scaled up for clinical use. Lastly, this study focused on ex vivo blood research, so in vivo performance may differ significantly. For example, leukocytes from bone marrow constantly enter the blood during CPB within the body, which was not the case in our experiments. Despite these limitations, our results justify further development and optimization of our novel disposable leukocyte remover to improve prognosis of patients after CPB.

Conclusion

Our novel disposable leukocyte remover can preferentially remove activated neutrophils from ex vivo blood of patients who underwent CPB, with minimal impact on other leukocyte subsets and blood components. These findings highlight the remover’s potential to reduce inflammation-related complications after bypass, justifying further investigation in larger studies.