Comparative effects of hemodialysis modalities on microparticle induction and neutrophil activation in a randomized cross-over study

Summary

Although the clearance of uremic toxins is comparable between expanded hemodialysis using a medium cut-off dialyzer (MCO+HDX) and online hemodiafiltration (OL-HDF), their effects on pro-inflammatory mediators have not been clarified. This study conducted a randomized cross-over study involving 12 thrice-weekly patients with hemodialysis who sequentially received MCO+HDX and OL-HDF with a washout period. Microparticles (MPs) and neutrophil activation were evaluated pre- and post-dialysis. OL-HDF showed elevated post-dialysis annexin V-positive MPs, predominantly from platelets and neutrophils. In vitro, post-dialysis MPs from both modalities aggravated human coronary artery endothelial cell (HCAEC) dysfunction. OL-HDF also enhanced post-dialysis neutrophil activation, as shown by elevated CD11b expression and plasma citrullinated histone H3 levels. These findings suggest that OL-HDF induces greater MP release and neutrophil activation than MCO+HDX, likely due to higher transmembrane pressures, which may contribute to endothelial damage and inflammation. The systemic impacts of HD modalities to induce both parameters might be clinically important.

Graphical abstract

Highlights

• •Expanded hemodialysis and online hemodiafiltration achieve comparable uremic toxin clearance
• •Both dialysis modalities differently influence pro-inflammatory mediator levels
• •Online hemodiafiltration triggers greater neutrophil activation and microparticle release
• •Results guide personalized dialysis strategies to minimize inflammation during treatment

Immunology; Cell biology

Introduction

Hemodialysis (HD) is an important method among several modalities of renal replacement therapy in patients with end-stage renal disease (ESRD), effectively to reduce serum concentrations of uremic toxins and correct fluid overload.^1,2 Because uremic toxin accumulation is one of the major causes of several complications of ESRD,^1,2 the removal of uremic toxins is one of the most important concerns. Indeed, uremic toxins consist of small-molecule (<500 Da), middle-molecule (>500 Da), including small middle-molecule (<25,000 Da) and large middle-molecule (≥25,000 Da), and protein-bound uremic toxins.^2,3 Unfortunately, both middle-molecule and protein-bound uremic toxins are correlated with cardiovascular complications and all-cause mortalities but are poorly removed by HD with a regular high-flux dialyzer.⁴ Then, the retention of large middle-molecule uremic toxins; for example, beta-2 microglobulin (β2M) with a molecular weight (MW) of 11,800 Da or other toxins with a higher MW, induces vascular calcification, elevates neutrophil function, and enhances mortalities in patients with chronic kidney disease (CKD).^5,6 Hence, several modalities aimed to removing high-MW toxins are introduced.

Accordingly, online hemodiafiltration (OL-HDF), based on the combined diffusion and convection mechanisms, provides a superior survival benefit over regular HD with a high-flux dialyzer.⁷ Due to the diminished uremic toxin diffusion clearance from hemodilution with pre-dilution OL-HDF and the heightened blood viscosity with more frequent dialyzer clots from hemoconcentration in post-dilution OL-HDF, the mixed-dilution OL-HDF (divided the pre- and post-dilutions in the same treatment) has been provided noninferior or even better efficacy when compared with post- and pre-dilution OL-HDF.³ In parallel, HD with a medium cut-off dialyzer (MCO+HDX) with larger pore sizes than a standard high-flux dialyzer also effectively removes middle-molecule and protein bound uremic toxins, similar to OL-HDF.³ While MCO+HDX can be performed in a regular HD machine that removes the toxins through the diffusion property, OL-HDF needs a specific machine for toxin removal through both convection and diffusion mechanisms. Thus, it is possible that the difference in intra-dialyzer between two HD modalities, such as transmembrane pressure, is higher in OL-HDF than in MCO+HDX, which might affect some circulating blood cells to produce pro-inflammatory mediators, which can be directly released into the patient’s circulation. This topic warrants careful consideration.

Indeed, cell derived microparticles (MPs), small cell membrane-derived vesicles with 0.1–1.0 μm in size formed by the outward blebbing of the cell membrane, are released from the activated cells or cell death. Although, the accumulation of uremic toxin in ESRD may cause systemic inflammation and cardiovascular complications in patients,^8,9 exposure to MPs might worsen these conditions.¹⁰ Indeed, the risk of cardiovascular mortality and overall mortality in patients with chronic HD is 10–20-folds and 100-folds, respectively, higher than in the general population.¹¹ Accordingly, MPs are well known as bioactive particles that can be involved in various mechanisms, including coagulation, pro-inflammation, and atherosclerosis, implying that the presence of MPs might worsen cardiovascular mortalities in ESRD.^12,13 During HD sessions, the shearing stress from direct contact between the patient’s cells and dialyzer membrane and/or during intradialytic hypotension produces MPs.¹⁴ While several studies demonstrate increased MPs after hemodialysis,^15,16 some report indicate a decrease in MP levels.¹⁷ In addition to generate MPs, the impact of HD on neutrophil activation has also been demonstrated.¹⁶ Neutrophils play a central role in inflammation and are thought to be major contributors to risks of cardiovascular complication in serveral diseases.^16,18,19 During early activation, neutrophils can release MPs, similar to platelets.¹⁶ Although there is a more effective removal of uremic toxins by OL-HDF and MCO+HDX over regular HD with a high flux dialyzer,³ HD-induced MP generation and neutrophil activation in OL-HDF and MCO+HDX might lead to some adverse effects. Because MPs and neutrophils from OL-HDF and MCO+HDX are never explored, and the possible different production of MPs and neutrophil activation might be another factor in the proper selection of hemodialysis modalities, a prospective randomized controlled cross-over design with in vitro experiments on a coronary arterial cell line was conducted.

Results

Demographic data and uremic toxin clearance

Among the 19 eligible patients, 14 consented to participate in the study. Two of these participants were excluded due to failure to complete the protocolized assessments, resulting in 12 patients who completed all study arms and were included in the final analysis (Figure 1). All participants tolerated both modalities of treatment. The baseline characteristics are detailed in Table 1, with a mean dialysis vintage of 9 ± 6.15 years, and the most common underlying disease is diabetic nephropathy. The characteristics of HD and the efficacy of uremic toxin removal are demonstrated in Table 2. There was a similar removal of the middle molecule uremic toxins, including β2M, free light chain (λ and κ), and alpha-1-microglobulin (α1M); however, albumin loss from MCO+HDX was higher than OL-HDF, and the transmembrane pressure of OL-HDF was higher than MCO+HDX (Table 2).

Figure 1.

The schema of the study design is demonstrated

OL-HDF, online hemodiafiltration; MCO+HDX, hemodialysis with median cut-off dialyzer.

Table 1.

Baseline characteristics

Parameter (mean ± SD) or n, (%)
Age, years	60 ± 6
Gender, n (% male)	6 (50%)
Dialysis vintage, years	9 ± 6
Dry weight, kg	56 ± 15
Kt/V	2.5 ± 0.5
Urea reduction ratio, %	86 ± 5
nPCR, g/kg/day	1.3 ± 0.3
AV fistula	8 (67%)
AV graft	4 (33%)
Unknown causes of CKD	6 (50%)
Diabetic nephropathy	4 (33.3%)
Chronic glomerulonephritis	2 (16.7%)
Hemoglobin, g/dL	12 ± 1
Platelet, 1,000/mm3	210 ± 522
Calcium, mg/dL	9.1 ± 0.6
Phosphate, mg/dL	4.2 ± 1.3
iPTH, pg/L	438 ± 444
Serum albumin, g/L	3.9 ± 0.2
β2M, mg/L	26 ± 6
α1M, mg/dL	14 ± 2
IS, mg/L	2 ± 1

nPCR, normalized protein catabolic rate; AV, arteriovenous; CKD, chronic kidney disease; IS, indoxyl sulfate.

Table 2.

Characteristics of the hemodialysis protocol

	MCO + HDX	OL-HDF
Characteristics of HD modalities
Dialysis machine	Fresenius 5008H	Fresenius 5008H
Dialyzer	Theranova 500	ELISIO 21H
Membrane	Polyether sulfone/polyvinylpyrrolidone	Polyether sulfone
Effective surface area, m2	2	2.1
Fiber inner diameter, um	180	200
Fiber wall thickness, um	35	40
The coefficient of ultrafiltration (KUF), mL/h/mmHg	59	82
Mass transfer area coefficient (KoA) urea, mL/min	1630	1976
Sieving coefficient of myoglobin	0.9	0.22
Sieving coefficient of albumin	0.008	0.002
Sterilization method	Stream	Dry gamma
Study treatment characteristics
Effective dialysis time, min	240 ± 0.08	242 ± 0.28
Actual blood flow rate, mL/min	396 ± 2.15	393 ± 9.45
Actual ultrafiltration volume, L	2.55 ± 0.62	2.77 ± 0.96
Transmembrane pressure, mmHg∗	27 ± 6.07	237 ± 22.45
Convection volume/session, mL	–	39.5 ± 4.12
Dialysate flow rate, mL/min	400	400
Effectiveness in removal of uremic toxin
The spKt/V urea	2.54 ± 0.68	2.56 ± 0.60
Urea RR, %	86.41 ± 4.84	86.41 ± 4.48
Indoxyl sulfate RR, %	60.49 ± 24.26	62 ± 19.86
β2M RR, %	82.57 ± 5.34	85.12 ± 3.87
κFLC RR, %	77.08 ± 10.62	77.65 ± 5.06
α1M RR, %	41.49 ± 11.46	30.13 ± 15.90
λFLC RR, %	50.81 ± 13.18	40.85 ± 13.92
Albumin Loss, g/session∗	3.51 ± 1.34	0.58 ± 0.54

spKt; single-pool urea Kt/V, RR; reduction ratio, β₂M; beta 2-microglobulin, κFLC; kappa free light chain, α₁M; alpha-1-microglobulin, λFLC; lamda free light chain. ∗; p < 0.05 between groups.

The induction of microparticles after hemodialysis

Firstly, PKH-26 (a lipophilic long-chain carbocyanine fluorescent dye for cell membrane labeling) was used to observe isolated pellets after differential centrifugation, and it was found that greater than 80% of isolated particles from the patient’s plasma were positive, suggesting there were MPs (data not shown). Following a previous publication, the measurement of MPs in plasma was performed by flow cytometry analysis using annexin V, the phospholipid-binding protein attached to phosphatidylserine (PS), a major mechanism responsible for MP production.²⁰ Additionally, several biomarkers were used to explore the possible original sources, including CD41 (platelets), CD66b (neutrophils), CD14 (monocytes), CD235a (red blood cells), and CD106 (endothelium).²¹ The representative flow cytometry analysis and gating methods for annexin V positive (annexin V+) MPs with other additional cell markers are demonstrated in Figure 2A. With the comparison between pre- and post-HD, annexin V+ MPs in post-HD with OL-HDF were higher than pre-HD, while the levels of MPs in pre-versus post-HD of MCO-HDX were not different (Figure 2B). In parallel, the levels of post-HD MPs from OL-HDF were also higher than those of MCO+HDX treatment, while the levels of pre-HD MPs of both HD methods were similar (Figure 2B). For the phenotypic analysis of MPs, the flow cytometry analysis indicated that MPs were produced more from platelets and neutrophils in post-HD, either with OL-HDF or MCO-HDX, while the abundance of MPs originating from RBC was reduced when compared with the MPs from the pre-HD session (Figures 2C and 2D). Notably, the majority of pre-HD MPs were produced from platelets, endothelial cells, and RBC, while post-HD MPs (both HD modalities) mostly originated from platelets and neutrophils, implying the impacts of the HD process on MP generation (Figures 2C and 2D). For the analysis of post-dialysis MPs, the MPs originated from platelets and neutrophils and demonstrated the most prominent MPs in all patients regardless of dialysis modes (OL-HDF or MCO-HDX), as demonstrated by a radar chart and a heatmap graph (Figures 2C–2E). With the analysis of delta change between pre- and post-HD, an increase in platelet MPs of OL-HDF was more prominent than MCO-HDX, while the alteration of MPs from other sources, including neutrophils, monocytes, epithelium, and RBC, was not different (Figure 2F).

Figure 2.

Flow-cytometric analysis of plasma microparticles (MPs) alteration during pre- and post-hemodialysis using mixed dilution online hemodiafiltration (OL-HDF) or hemodialysis with median cut-off dialyzer (MCO+HDX)

The representative schema of flow-cytometric analysis for measuring plasma MPs (A) and characteristics of annexin V positive MPs (annexin V+ MPs) during the pre- and post-dialysis using OL-HDF or MCO+HDX (B) are demonstrated (n = 12/group). Analysis of the source of cells that produce MPs from the pre-dialysis (orange color) and postdialysis (blue color) using OL-HDF or MCO+HDX as indicated by the radar charts (C and D; the unit of C and D is x102 particles/μL), heatmap of delta change of MPs between pre-versus post-dialysis MPs (the color codes represent the abundance) (E), and a dot-plot graph (F) are demonstrated (n = 12/group). ∗, p < 0.05 vs. the indicated groups.

Effect of hemodialysis induced microparticles to promote human coronary artery endothelial cell dysfunction

Endothelial cell dysfunction is a prior mechanism of cardiovascular diseases (CVDs).²² To establish the effect of HD-induced MPs on prime cardiovascular complications in patients with chronic HD, in vitro experiments on primary human coronary artery endothelial cells (HCAECs) were examined. The PMA stimulation was also applied to simulate the provoking environment.²³ Culture of HCAECs with post-HD MPs provided that CFSE-labeled post-HD MPs (green events) were uptake by HCAECs in a time-dependent manner observed by fluorescent imaging (Figures 3A and 3B). This study assessed HCAEC dysfunction by characterizations of adhesion molecule alteration, tight junction protein expression, and cell apoptosis.²⁴ The results showed that both post-HD MPs can induce cell dysfunctions via an increase in VCAM-1 expression, an adhesion molecule to enhance WBC attachment,²⁵ and cell apoptosis in a dose dependent manner (Figures 3C and 3D). As well, the results of tight junction observation also exhibited that both post-HD MPs interfere with ZO-1 tight junction formation in a time dependent manner (Figure 3). Moreover, the results of cell proliferation, detected by the MTS assay, showed that post-HD MPs can slightly suppress cell proliferation (Figure S1A). Compared to pre-HD MPs, post-HD MPs from both HD processes demonstrated a greater effect in enhancing VCAM-1 expression along with a higher trend, but not significant, in cell apoptosis (Figures S1B and S1C). The different capability between post-HD MPs from the MCO+HDX session and the mixed OL-HDF session was not observed. These findings may have suggested that the altered MP phenomenon induced by both HD modalities may prime endothelial cell dysfunction in a dose dependent manner.

Figure 3.

Effects of post-hemodialysis microparticles from the mixed dilution online-HDF (OL-HDF) or hemodialysis with median cut-off dialyzer (MCO-HDX) on human coronary artery endothelial cells (HCAECs)

The representation of carboxyfluorescein succinimidyl ester (CSFE)-stained MPs (green color) represented by CSFE fluorescent intensity with the representative immunofluorescent pictures (A and B) are demonstrated (n = 6/group, scale bar = 10 $\mu$m). The red color in A is zona occluden-1 (ZO-1; a tight junction molecule) and the blue color is DNA-stained by 4′,6-diamidino-2-phenylindole (DAPI). The influences of the post-dialysis MPs from OL-HDF (C) or MCO+HDX (D) with 1x10⁴ (light color) or 1x10⁵ (heavy color) particles/μL at 24 h after inoculation, compared with control (only media but not MPs) in the condition with or without phorbol myristate acetate (PMA, the positive control activator) on HCAECs as indicated by vascular cell adhesion molecule 1 (VCAM-1) and cell apoptosis, are shown. The effect of MPs on tight junction molecules ZO-1 is also demonstrated through the intensity score of the red color with representative immunofluorescent (E, F) (n = 6/group, scale bar = 10 $\mu$m). Notably, the MPs were not stained to avoid the noise from the green-fluorescent color. ∗p < 0.05 vs. control within group; #p < 0.05 vs. same condition between groups.

Post-hemodialysis neutrophil activation

While post-dialysis MPs originated from platelets might be due to the interaction between platelets and the dialyzer membrane,²⁶ post-dialysis neutrophilic MPs possibly indirectly indicate the impact of post-HD neutrophil activation. The isolated neutrophils from post-HD with either OL-HDF or MOC-HDX as evaluated by flow cytometry analysis exhibited an increased expression of CD11b, a marker of granulocyte activation (used for cell migration, adhesion, transmigration across blood vessels, and cell-cell or cell-matrix interaction) (Figures 4A and 4B). Additionally, the post-HD neutrophils of OL-HDF, but not MCO-HDX, demonstrated the higher neutrophil extracellular traps (NETs), as indicated by increased citrullinated histone 3 (citH3) in plasma, but not by the morphology of nuclei using DAPI staining (Figures 4C–4E). Because the post-dialysis neutrophils might be more susceptible to NETs, a standard NET inducer (PMA) was used to test the neutrophils. Accordingly, PMA similarly activated NETs in neutrophils from either pre- or post-dialysis in both dialysis modalities (Figure 4D). In other phenotypes, a trend toward a slight increase in CD66b and CD63 (Figure 4F), both of which represent degranulation activity, was also detected, but significant alteration of plasma pro-inflammatory cytokines, including TNF-a, was not detected in both MCO+HDX and mixed OL-HDF (Figure 4G). These findings may have suggested that both HD modalities, especially OL-HDF, provide a priming effect to activate neutrophils, leading to an enhanced NETs of neutrophils in response to stimulators.

Figure 4.

Impact of hemodialysis using mixed dilution online hemodiafiltration (OL-HDF) or hemodialysis with median cut-off dialyzer (MCO+HDX) to activate neutrophils

The representative schema of flow-cytometric analysis for measuring CD11b expression on isolated neutrophils (A) during the pre- and post-hemodialysis using OL-HDF or MCO+HDX (B) are demonstrated (n = 6/group). Neutrophil extracellular traps (NETs) in blood at pre-and post-hemodialysis using OL-HDF or MCO-HDX, as indicated by the percentage of NETs with representative fluorescent pictures after the stimulation by phorbol myristate acetate (PMA, the positive control activator) and control (unstimulated) (C and D) (scale bar = 10 $\mu$m), the citrullinated histone H3 (cit-H3) level (E), the expression of CD66b and CD63 (F,G), and the tumor necrosis factor alpha (TNFa) production (H) are demonstrated (n = 6/group). ∗, p < 0.05 vs. the indicated groups.

Discussion

Because of the excellent removal of small-molecule uremic toxins in most of the current HD modalities, the removal of middle-molecule and protein-bound toxins is increasingly concerned, which is similar between the mixed dilution online HDF (OL-HDF) and expanded HD with a median cut-off dialyzer (MCO+HDX).³ Hence, MCO+HDX might be a clinically valuable option for patients with HD when OL-HDF is not available.²⁷ Additionally, MCO+HDX (the dialyzer with larger pore sizes than a standard high-flux dialyzer) might be easier to use with a reduced HD cost in a clinical situation with only the standard HD machine without the part for the online system.^11,27 However, the effect of triggering inflammation between both HD modalities has never been compared. This current study demonstrated an increase in platelet- and neutrophil-derived MPs after the HD session, and the post-HD MPs in OL-HDF were higher than in MCO-HDX, parallel to the relative increase in NET formation, both of which will be sequentially discussed.

The induction of post-hemodialysis microparticles and neutrophil extracellular traps

The cardiovascular mortality in patients with chronic HD is approximately 100-folds higher than that in the general population¹¹ which, at least in part, is due to (i) the direct impact of some uremic toxins and (ii) the systemic inflammation from ESRD or dialysis modalities. Indeed, some of the uremic toxins, for example, indoxyl sulfate, p-cresol, and trimethylamine-N-Oxide (TMAO), can directly induce atherosclerosis.^28,29 Meanwhile, the chronic systemic inflammation from ESRD due to increased oxidative stress or the translocation of microbial molecules from the gut into the blood circulation (leaky gut) might also elevate cardiovascular risks.^30,31,32,33 Hence, the effective removal of uremic toxins (both small and middle molecules) and oxidative stress using OL-HDF or MCO+HDX is one of the strategies to reduce atherosclerosis risk.³⁴ However, the HD process involving blood contact with “foreign” surfaces may also be a potential source of inflammation.³⁵ For these reasons, recent dialysis research has been aimed not only at developing procedures for inflammatory toxin removal but also at improving bio-incompatibility, which is a potential inflammatory inducer of the extracorporeal circuit.^35,36 Blood contact with the foreign surface promotes a variety of complex and interrelated events during the HD process, such as the activation of complement and coagulation,³⁷ which might directly induce MP production and neutrophil activation, as demonstrated by the in vitro activation of C5b-9 (a complement protein complex).^38,39 In the pre-dialysis condition, the annexin V-positive MPs mainly originated from platelets, endothelial cells, and RBCs, consistent with several previous studies^38,39 that supported uremia-induced activation of these cells³⁹ and some parts of these apoptotic cells might produce MPs as mentioned in apoptosis-induced MPs.⁴⁰ However, in the post-dialysis condition, the increased annexin V-positive MPs mainly originated from platelets and neutrophils, implying the activation of these cells partly by the HD procedure in parallel with demonstrated post-HD neutrophil activation through the elevation of CD11b in both OL-HDF and MCO-HDX. The repetitive mechanical stress between the cells and dialyzer membrane can activate both platelets and neutrophils and induce MP generation, referred to as mechano-transduction-related hemostasis and neutrophil functions.^41,42 The transmembrane pressure might be correlated with the shearing stress, as the transmembrane pressure of OL-HDF was higher than that of MCO-HDX, in parallel with the higher MPs and plasma cit-H3 (a biomarker of NETs) in OL-HDF compared with MCO-HDX. Another factor that might be correlated with the higher MPs in OL-HDF compared with MCO-HDX might be due to the different material of the dialyzer membranes. However, polyether sulfone, the main material for dialyzers of both modalities, is a biocompatible material. Of note, the endothelial MPs were not the major component of MPs in our study, possibly due to the controllable level of uremia that safeguarded the uremia-induced endothelial MPs.⁴³ Indeed, increased post-HD MPs (both the phenomenon and the levels) and neutrophil activations might be potential factors which can be contributed to the elevation of systemic inflammation and cardiovascular risks through several mechanisms.⁴⁴ These findings suggest that the different properties of HD modes extend beyond the removal of uremic toxins.

Post-dialysis microparticles: a concern for the selection of hemodialysis modalities

Although elevated levels of MPs have previously been associated with endothelial dysfunction and may be used to predict cardiovascular death in patients with ESRD,^14,17 evidence which suggests the direct impact of HD-induced MPs on the cardiovascular system has never been demonstrated. Here, the endocytosis of MPs by primary human coronary artery endothelial cells (HCAECs) at 2 h post-incubation in a time dependent manner caused endothelial dysfunction, as indicated by VCAM-1 upregulation and endothelial tight junction injury, which might be correlated with cardiovascular diseases.^45,46 Due to the comparable MP concentration between OL-HDF and MCO+HDX, MP-induced endothelium dysfunction caused by both HD modalities might be similar. Thus, the quantification of MPs in patients with HD might be another factor that should be considered. Although the downstream mechanisms of endothelial damage by MPs were not explored here due to the limited scope of the study and the proof-of-concept purpose, platelet- and neutrophil-MPs might activate endothelial cells in the same manner as the intact platelets and neutrophils.^8,47 Indeed, both platelets and neutrophils are important components of endothelial dysfunction, which leads to atherosclerosis.^8,10,47 Besides the cardiovascular risks, MPs and NETs also play crucial roles in agonistic coagulators, which is an important problem in critical renal replacement therapy (CRRT).^15,48,49 Hence, the selection of dialysis modalities might not only rely on the effective clearance of uremic toxins but also on the generation of MPs, NETs, and systemic inflammation. This is the first study to compare the acute effect of HD modalities on the induction of MPs. The higher transmembrane pressure of OL-HDF might be an important cause of MP generation. Therefore, the control of transmembrane pressure might be an important strategy to reduce MP generation and systemic inflammation.⁵⁰ Moreover, the production of MPs after HD might also be more prominent in other modalities with high transmembrane pressure; for example, post-dilution OL-HDF and HD with a low-flux dialyzer.⁵¹ Meanwhile, the MCO+HDX seems to be an interesting HD modality which can be performed by most of the regular HD machines with an acceptable clearance of middle molecules and protein-bound uremic toxins along with minimum inductions of MPs and NETs. Although MCO-HDX demonstrated a decrease in MP induction and neutrophil activation compared to OL-HDF, the 6-fold increase of albumin loss per session in MCO-HDX raises concerns about long-term nutritional consequences, particularly in vulnerable patient populations,⁵² which may be a consideration when choosing hemodialysis modalities. More studies to demonstrate MPs in other dialysis modalities and the clinical impacts of HD induce MPs on patients with chronic HD are interesting.

Conclusion

In conclusion, the more prominent post-HD phosphatidylserine exposing MPs of mixed dilution OL-HDF over MCO-HDX, mainly due to the higher transmembrane pressure, were demonstrated. The post-HD MPs were primarily derived from platelets and neutrophils, which might directly activate and damage coronary arterial cells, leading to cardiovascular complications. Moreover, the induction of inflammation, represented by an increase in surrogate NET biomarkers, was also found in mixed dilutions of OL-HDF. Not only the clearance of uremic toxins, but also the MP production, inflammatory induction properties, and albumin loss, should be considered in the selection of hemodialysis modalities. More studies are warranted.

Limitations of this study

The number of patients was quite small due to the single center observation. Additionally, the participants were mainly elderly, which may be linked to altered immune responses or comorbidities that could affect inflammation, vascular function, and albumin metabolism, thereby influencing microparticle production and neutrophil activation. Furthermore, the 2-week duration of HD treatment might not be long enough to assess various clinical outcomes, especially cardiovascular complications. Future studies with larger numbers of participants and longer-term follow-up with various HD modalities are crucially required. Also, the MP production in other HD modalities, including conventional HD with low-flux or high-flux dialyzers, was not evaluated. Urea and indoxyl sulfate were chosen as representative middle-molecule and protein-bound uremic toxins due to their clinical significance and the availability of the assays. However, one cannot assume that both HD modalities can remove all these categories of toxins. Although the removal kinetics of uremic toxins tend to correlate within each molecular class,⁵³ further studies are required to generalize this finding to all toxins in these categories. Moreover, other parameters represented systemic inflammation such as cell-free DNA,⁵⁴ or interleukin-6 (IL-6)⁵⁵ were not performed.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fullfilled by the lead contact, Asada Leelahavanichkul (aleelahavanit@gmail.com).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •All data reported in this article will be shared by the lead contact upon request.
• •This article does not report original code.
• •Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Star★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
FITC conjugated anti-CD235a antibody	Biolegend	Cat#349104
FITC conjugated anti-CD66b antibody	BD bioscience	Cat#555724
FITC conjugated anti-CD106 antibody	BD bioscience	Cat#551146
PE conjugated anti-CD63 antibody	BD bioscience	Cat#556020
PE-cy5 conjugated anti-CD41 antibody	Biolegend	Cat#303708
PerCP conjugated anti-CD14 antibody	Biolegend	Cat#325632
APC-cy7 conjugated anti-CD11b antibody	BD bioscience	Cat#557754
APC conjugated annexin V	Biolegend	Cat#640920
Anti-citrulline Histone 3 (citH3) antibody	Abcam	AB5103
Anti-citrulline Histone 3 (citH3) antibody	Abcam	AB212082
Anti-zonula occludens-1 (ZO-1) antibody	Invitrogen	Cat#61-7300
Alexa Fluor™488 conjugated anti rabbit antibody	Invitrogen	Cat#A-11034
Alexa Fluor™647 conjugated anti rabbit antibody	Invitrogen	Cat#A-21245
Biological samples
Blood from participants with HD	King Memorial Chulalongkorn Hospital (KMCH)	IRB 322/63 TCTR 20201219001
Chemicals, peptides, and recombinant proteins
RPMI medium 1640	Gibco	Ref. 11875-093
Fetal bovine serum	Gibco	Ref.A5256701
D-PBS	Hyclone™	Cat#SH30028.02
HBSS	Hyclone™	Cat#SH30588.02
RBC lysis buffer	Biolegend	Cat#420302
Ploymorph prep solution	Serumwerk Bernburg	Prod.no.1895
Tryphan blue	invitrogen	Ref.T10282
PMA (Phorbol 12-myristate 13-acetate)	Thermo Scientific Chemicals	Cat#J63916.LB0
5(6) CFDA, SE (5-(and-6)-Carboxyfluorescein Diacetate, Succinimidyl Ester)	invitrogen	Ref.qfgV12883
PKH26	Sigma-Aldrich	UNSPSC Code 12352207
DAPI	Thermo Scientific™	Cat#62248
Staining buffer	BD bioscience	Cat#554656
4.2% Paraformaldehyde	BD bioscience	Cat#554655
Trypsin-EDTA for Primary Cells	ATCC	PCS-999-003 Lot No. 81226888
Trypsin Neutralizing Solution	ATCC	PCS-999-004 Lot No. 80916444
Critical commercial assays
Theranova 500	Terumo	KCMH laboratory
ELISIO 21H	Fresenius Medical Care	KCMH laboratory
Cobas c502 analyzer	Roche Diagnostics	KCMH laboratory
Alliance 2695 HPLC	Waters	KCMH laboratory
α1-microglobulin ELISA kit	Abcam	AB108884
Tumor necrosis factor ELISA kit	Invitrogen	Ref. 88-7346-22
3.0 μm Latex beads, polystyrene	Merck	UNSPSC Code 41121800
Flow-Count Fluorosphere	Beckman Coulter	Ref. 7547053
Experimental models: Cell lines
Primary Human Coronary Endothelial Cells (HCAECs), Female	ATCC	PCS-100-020 Lot No. 70001400
Endothelial Cell Growth Kit-VEGF	ATCC	PCS-100-041 Lot No. 80524190
Vascular Cell Basal Medium	ATCC	PCS-100-030 Lot No. 80827212
Software and algorithms
SPSS statistical package version 22	IBM-SPSS	https://www.ibm.com/products/spss
GraphPad Prism version 9	SAS Institute	https://www.graphpad.com/features
Microsoft Excel 2019	Microsoft	https://www.microsoft.com/th-th/microsoft-2019/excel
Flowjo software version 10	Flowjo, inc	https://www.flowjo.com/

Experimental model and study participant details

The study was approved by the Ethical Committee, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand (IRB No. 322/63) with Thai Clinical Trials Registry (TCTR 20201219001) registration. The single-center prospective cross-over randomized controlled trial (RCT) was conducted at the division of nephrology, King Chulalongkorn Memorial Hospital (KCMH), Bangkok, Thailand, with Informed consent obtained from all patients. The inclusion criteria were stable chronic hemodialysis (HD) with thrice-a-week for at least 6 months with an age more than 18 years old, adequate small-molecule uremic toxin removal (spKt/V urea greater than 1.2/HD session), residual diuresis below 100 mL/day, blood flow rate (BFR) during dialysis of more than 400 mL/min, and stable hemodynamics for more than 2 weeks before enrollment. The exclusion criteria were active infection, symptomatic cardiovascular disease, advanced stage malignancy, and decompensated cirrhosis. The schema of the experimental design was indicated in Figure 1. Briefly, there was a 2-week run-in period using high-volume post-dilution OL-HDF with high-flux dialyzer (the standard treatment in the KCMH). Subsequently, all patients were randomly allocated by blocked randomization into MCO+HDX or mixed dilution OL-HDF (4 h) for 2 weeks before a 2-week wash-out phase and cross-over. The Fresenius 5008H (Fresenius Medical Care, Germany) HDF machine with ELISIO21H high-flux dialyzer for mixed OL-HDF and with Theranova (Gambro Dialysatoren GmbH, Germany) for MCO+HDX thrice-a-week with 400 mL/min of both BFR and dialysate flow rate (DFR) and standard unfractionated heparin. The purity of the dialysis fluid met the standards of ultrapure water (ISO criteria). The removal of toxins was determined by the reduction ratio (RR) of small molecules (urea RR), middle-molecules using RR of β2 microglobulin (β2M), free light chain (FLC) (both κ and λ), and alpha-1 microglobulin (α1M), and protein-bound uremic toxins (indoxyl sulfate RR) following a previous publication.³

Method details

Data and sample collection

Baseline demographic data were obtained at enrollment. Laboratory parameters were averaged from values collected at the end of each 2-week treatment phase before and after crossover. Blood samples were collected in recommended anticoagulant tubes (depending on the assay) immediately before and within 1 h after HD sessions. Plasma and cells were immediately separated after collection for downstream analyses as described below.

Uremic toxin clearances

The concentrations of these uremic toxins were measured in the plasma of the patients in pre- and post-hemodialysis with several assays, as following: (i) nephelometry (the central laboratory of the King Chulalongkorn Memorial Hospital with Cobas c502, Roche Diagnostics, Basel, Switzerland) for β2M (MW 12 kDa), kappa-free light chain (κFLC; MW 22 kDa), lambda-free light chain (λFLC, MW 45 kDa), and albumin (MW 66 kDa) (ii) ELISA kit (AB108884) (Abcam, Cambridge, UK) for α1M (MW 33 kDa), (iii) high-performance liquid chromatography (HPLC Alliance 2695; Waters, Zellik, Belgium) for indoxyl sulfate (IS, MW 213 Da) and urea (MW 60 Da). The RR was calculated by dividing the pre- and post-dialytic blood levels of the sessions by the following equation: RR (%) = [1-(cCpost/Cpre)]x100, where Cpre and Cpost refer to the concentration of the molecules in the pre- and post-HD sessions, respectively, and cCpost is the corrected Cpost due to the weight reduction during the hemodialysis session. The cCpost was calculated by the equation: cCpost = Cpost/[1+(ΔBW/0.2BWpost)]; where ΔBW is the weight lost during the HD session and BWpost is the body weight at the end of the HD session. Baseline demographic data were collected at the enrollment, and laboratory parameters were the average of the values at the end of sessions (2-week-period before and after the cross-over).

Microparticle isolation and characteristics

The heparinized plasma was collected pre- and post-HD within 1 h for MP measurement following previous studies.^3,17,21,56 Briefly, the plasma was centrifuged at 4°C for 10 min at 5,000 g to remove debris and apoptotic bodies and re-centrifuged at 15,000 g for 60 min at 4°C to precipitate the MPs, which were then washed and resuspended in 500 μL of Hank’s Balanced Salt Solution (HBSS). Then, the isolated MPs were labeled with allophycocyanin (APC) conjugated annexin V, fluorescein isothiocyanate (FITC) conjugated anti-CD66b (neutrophil marker) (BD Biosciences, NJ, USA), R-phycoerythrin with Cyanine (PE-cy5) anti-CD41 (platelet marker) (Biolegend, CA, USA), FITC conjugated anti-CD106 (endothelial cell marker) (BD Biosciences, NJ, USA), peridinin-Chlorophyll-Protein (PerCP) conjugated anti-CD14 (monocyte marker) (Biolegend, CA, USA), FITC conjugated anti CD235a (RBC marker), and PE conjugated anti- CD142 (tissue factor) (BD Biosciences, NJ, US). The MP population was gated by comparison with calibrant beads of 3 μm in diameter, and the absolute number of MPs was calculated using counting beads (Beckman Coulter, CA, USA).

Endothelial cell culture and stimulation

The experiment was based on previous studies.^21,57 Briefly, human coronary artery endothelial cells (HCAECs) were cultured in a confluent monolayer using endothelial cell growth medium before adding post-dialysis MPs (1 × 10⁵ particles/uL) and incubated at 37°C, 5% CO2. The internalization of MPs and cell tight junction analysis were examined by immune-fluorescence microscopy. Briefly, MPs were labeled with 0.5 μM of carboxyfluorescein succinimidyl ester (CFSE), green-fluorescent color (ThermoFisher Scientific, MA, USA) for 5 min at room temperature before being treated using 1 × 10⁵ MPs/uL in the monolayered HCAECs at 37°C, 5% CO2. Then, non-adherent CFSE-MPs were removed by gently washing 3 times with HBSS before staining with Alexa Fluor647 conjugated antibody against Zonula occludens-1 (ZO-1; the cell tight junction molecule) (Thermo Fisher Scientific, MA, USA) for 60 min at room temperature. The slides were washed and mounted using mounting buffer containing 4′,6-diamidino-2-phenylindole (DAPI, a fluorescent color for the DNA staining) (Thermo Fisher Scientific, MA, USA). The HCAEC dysfunctions after being activated by MPs were determined by flow cytometry using apoptosis kits (BD Biosciences, NJ, USA) for cell death and VCAM-1 expression (BD biosciences, NJ, USA) for adhesion molecule expression. The proliferation were evaluated from treated HCAECs compared to untreat controls by MTS cell proliferation assay kit (Promega, WI, USA) according to the manufacturer’s recommendation.

Neutrophil isolation and neutrophil extracellular traps

Because MPs and the dialysis procedures might activate neutrophils, neutrophil activation and NETs were measured from the collected blood sample according to previous protocols.^58,59,60 Briefly, neutrophils from the collected blood were immediately separated by density gradient centrifugation (Ficoll-Polymorprep centrifugation; Serumwerk, Bernburg, Germany) using 800 g centrifugation for 30 min at room temperature before red blood cell (RBC) lysis by the lysis buffer (Biolegend, CA, USA), and washed twice with deionized phosphate buffer solution (DPBS) (Gibco, MA, USA). Then, the isolated neutrophils were maintained in Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco, MA, USA) containing 10% fetal bovine serum (FBS) (Gibco, MA, USA) at 37°C with 5% CO2. For the analysis, the isolated neutrophil was stained with APC-cy7 conjugated anti-CD11b antibody (dilution 1:50) (BD Biosciences, NJ, US) and measured by flow cytometric analysis. In parallel, the isolated neutrophil was stimulated with phorbol myristate acetate (PMA), an inducer for neutrophil extracellular traps (NETs) at 100 nM (MERCK, Darmstadt, Germany) for 120 min at 37°C with 5% CO2, before staining with DAPI for demonstrating NET formation through the nuclear morphology and anti-citrulline Histone 3 (citH3) antibody (Abcam, Cambridge, UK), a direct biomarker for NETs⁵⁸ and Alexa Fluor488 conjugated anti rabbit antibody (Abcam, Cambridge, UK). The NET formation was observed by immunofluorescence staining on a web-like structure positive with DAPI and citH3 expression using a fluorescent microscope (Olympus, Japan) with 10 representative pictures from each slide.

Quantification and statistical analysis

Descriptive statistics included mean with standard deviation (SD) and median with interquartile range (IQR) values for continuous variables and percentage for categorical variables. Paired T-test or Wilcoxon test were used to comparisons of the pre- and post-hemodialysis values. Analysis of variance (ANOVA) or Kruskal-Wallis test with multiple comparisons were applied for comparisons among multiple groups. two-way ANOVA was used for repeated measures. The model was fit by restricted maximum likelihood to provide an unbiased estimation of the variance and covariance parameters. The treatment and sequential effects were the result of the crossover design. All analyses were measured using a FACS LSR II cytometer (BD Biosciences, Franklin Lakes, NJ, USA) with the FlowJo V10 (Ashland, DE, USA). Two statistical software packages, including Graphpad Prism version 9 (SAS Institute, US) and IBM-SPSS version 22 (IBM-SPSS, US), were used for data visualization and statistical analyses, respectively. Statistical significance was considered when the p-value <0.05.

Acknowledgments

The authors gratefully thank all participating nephrologists and the nurses at the KCMH hemodialysis unit for any data and/or materials made available for the present study. Furthermore, we thank Medical Research Center (MRC), Faculty of Medicine, Chulalongkorn University for research services. This research was funded by the National Research Council of Thailand (NRCT) (N84A680764, N32A680647, and N42A680063), Ratchadapiseksompotch Fund, Faculty of Medicine, Chulalongkorn University (RA-MF-36/68). The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Author contributions

Conceptualization, A.C., J.E., K.S-k., and A.L.; data curation, A.C., J.E., and K.T.; formal analysis, A.C., K.S-k., and A.L.; investigation, A.C., K.S-k., and J.E.; project administration, A.C., J.E., and A.L.; methodology, J.E., S.U., K.T., and A.L.; software, A.C. and K.S-k.; supervision, A.C., S.U., K.T., and A.L.; validation, A.C., J.E., S.U., K.T., and A.L.; visualization, A.C. and K.S-k.; writing – original draft, A.C., J.E., K.S-k., and A.L.; writing – review and editing, S.U., K.T., and A.L. All authors critically revised and approved the final version of the article.

Declaration of interests

All authors approved the submission of the final article. The authors have disclosed that they do not have any potential conflicts of interest.

Published: September 3, 2025