Modality matters: Effect of dialysis modality on soluble intercellular adhesion molecule-1 in children on regular hemodialysis

Abstract

BACKGROUND

Intercellular adhesion molecule-1 (ICAM-1) plays a key role in mediating leukocyte adhesion and endothelial injury. Elevated circulating levels of soluble ICAM-1 (sICAM-1) in patients undergoing hemodialysis (HD) have been linked to a higher incidence of cardiovascular complications.

AIM

To evaluate the impact of dialysis modality and membrane performance on post-dialysis sICAM-1 levels in children undergoing maintenance HD.

METHODS

This study included 26 children on low-flux HD for at least 3 months and 26 sex-matched and age-matched healthy controls. Patients were shifted to high-flux HD for 3 months, then subsequently to post-dilutional online hemodiafiltration (OL-HDF) for additional 3 months. At the end of each study phase, blood samples were obtained before and after a mid-week HD session for sICAM-1 measurement. To account for hemoconcentration, post-session ICAM-1 were hematocrit-adjusted using van Beaumont’s formula yielding corrected ICAM-1 (cICAM-1).

RESULTS

Circulating post-dialysis cICAM-1 levels increased significantly after low-flux HD compared with the corresponding pre-dialysis levels (P = 0.002), and were also significantly higher than those of the control group (P < 0.001). Following conversion to high-flux HD, post-dialysis cICAM-1 concentrations decreased significantly compared to pre-dialysis values (P < 0.001), although they remained elevated compared to controls (P = 0.026). In contrast, OL-HDF was associated with a substantial decline in post-session cICAM-1 levels (P < 0.001), yielding values comparable to those observed in the control group (P = 0.898).

CONCLUSION

Children on maintenance HD exhibit elevated sICAM-1 levels. Our findings suggest that higher-efficiency dialysis modalities, such as OL-HDF, may attenuate dialysis-related inflammation and endothelial activation in these patients.

Key Words: Soluble intercellular adhesion molecule-1; Hemodialysis; Online hemodiafiltration; Chronic kidney disease; Low-flux hemodialysis; High-flux hemodialysis

Core Tip: Children on chronic hemodialysis (HD) are exposed to persistent inflammation, which contributes to endothelial activation and cardiovascular risk. The type of dialysis modality may influence this inflammatory burden. This study evaluates the effect of low-flux HD, high-flux HD, and online hemodiafiltration (OL-HDF) on circulating intercellular adhesion molecule-1 (ICAM-1), a marker of endothelial activation, in pediatric patients. Low-flux HD increased post-session soluble ICAM-1 levels, suggesting dialysis-related endothelial injury, whereas high-flux HD and OL-HDF significantly reduced its levels, with OL-HDF normalizing ICAM-1 to values comparable to healthy controls. These findings suggest that selecting more biocompatible and high-efficiency dialysis modalities may help mitigate dialysis-related inflammation and potentially reduce long-term endothelial activation in pediatric HD patients.

INTRODUCTION

Chronic kidney disease (CKD) patients experience profound disturbances in immune system regulation involving both innate and adaptive immune responses. This may be manifested as either immune suppression or persistent immune activation resulting in “inflammaging”, a chronic systemic state marked by ongoing inflammation, a pro-coagulant environment, activation of leukocytes, and increased circulating pro-inflammatory cytokines[1,2]. Cytokine activation in CKD has been shown to markedly upregulate the expression of intercellular adhesion molecule-1 (ICAM-1) on endothelial, epithelial, and immune cells[3]. ICAM-1 is a membrane-bound glycoprotein belonging to the immunoglobulin superfamily that plays a crucial role in mediating interactions between leukocytes and the vascular endothelium. Through these interactions, ICAM-1 promotes the adhesion of monocytes, lymphocytes, and neutrophils to activated endothelial cells and facilitates their migration from the bloodstream into inflamed tissues[4]. Elevated circulating levels of the soluble ICAM-1 (sICAM-1) have been reported in patients undergoing hemodialysis (HD). This increase may occur as a consequence of both reduced renal clearance and increased production in response to inflammatory stimuli and endothelial dysfunction[5]. Patients with CKD are exposed to increased oxidative stress and accumulation of uremic toxins that collectively accelerate the development of atherosclerosis. Expression of ICAM-1 may also be stimulated by several pathological conditions such as hemodynamic stress, hypertension, systemic vasculitis, diabetes mellitus, infection, endotoxemia, and active thrombosis[5]. The inflammatory burden is further exacerbated in such patients due to several extracorporeal treatment-related factors, including dialysis water impurities, inadequate microbiological quality of dialysate, bio-incompatibility of dialyzer membranes or extracorporeal circuits, and complications related to vascular access such as infection or thrombosis. Together, these factors may augment systemic inflammation and contribute to the increased cardiovascular risk observed in dialysis populations[6,7]. Conventional HD cannot fully reproduce the physiological functions of native kidneys because solute removal occurs intermittently rather than continuously and follows a non-physiological pattern. Moreover, its ability to eliminate uremic toxins of varying molecular sizes remains limited, contributing to persistent dialysis-related complications, particularly chronic inflammation.

Improvement in the effectiveness of HD can be achieved through several approaches. These include increasing both blood flow and dialysate flow rates, using dialyzers with larger membrane surface areas, ensuring optimal purity of dialysis water, and utilizing high-flux dialysis membranes capable of removing middle-molecular-weight solutes[8,9].

Online hemodiafiltration (OL-HDF) represents a more advanced modality of kidney replacement therapy that combines diffusive and convective mechanisms of solute transport. This technique has been proposed to offer clinical advantages over conventional HD by enhancing the removal of middle-molecular-weight toxins and reducing dialysis-associated inflammatory responses, particularly when high convection volumes, ultrapure dialysate, and high-flux membranes are used[10]. In pediatric dialysis populations, Fischbach et al[11,12] demonstrated that daily OL-HDF was associated with substantial improvement in cardiovascular risk profiles and enhanced growth outcomes among children receiving chronic dialysis therapy. Despite these encouraging findings, the influence of dialysis modality and dialyzer membrane characteristics on post-dialysis sICAM-1 levels in children remains insufficiently explored.

A clearer understanding of these relationships may provide valuable insights into the mechanisms underlying dialysis-associated inflammation and cardiovascular risk in pediatric CKD. Therefore, the present study aimed to investigate whether dialyzer membrane performance and dialysis modality influence post-dialysis circulating sICAM-1 levels in children receiving maintenance HD.

MATERIALS AND METHODS

Study population

This prospective study included 26 children aged 16 years or younger with incident end-stage kidney disease (ESKD) who had been clinically stable on conventional low-flux HD for a minimum of three months at the Pediatric Dialysis Unit of the Children’s Hospital, Ain Shams University. Twenty-six healthy individuals matched for age and sex were recruited as controls. Ethical approval for the study was obtained from the Institutional Review Board of Ain Shams University Hospital (Approval No. FMASU R124/2024). All procedures involving human participants were carried out in compliance with the ethical principles established by the Declaration of Helsinki.

Children were excluded if they had received HD for less than three months or had evidence of active infection, infective endocarditis, viral hepatitis, active thrombosis, or autoimmune disease. Additional exclusion criteria included acute kidney injury, diabetes mellitus, cardiomyopathy, malignancy, recent blood transfusion within the preceding month, insufficient vascular access blood flow, failed renal transplantation, or ongoing immunosuppressive therapy.

Study phases

At baseline, all participants underwent a comprehensive clinical evaluation, including the collection of demographic data, the underlying cause of CKD, and dialysis duration. A detailed physical examination was performed to exclude signs of infection, inflammation, or thrombotic complications. Baseline laboratory investigations included blood sampling obtained immediately before and after a mid-week HD session. Subsequently, patients were transitioned to high-flux HD for three months, followed by post-dilution OL-HDF for an additional three months.

At the end of each treatment phase, laboratory measurements were repeated before and after the mid-week dialysis session, as illustrated in Figure 1. Because children with ESKD require uninterrupted renal replacement therapy, a washout period between dialysis modalities was not feasible. To minimize potential carryover effects, each dialysis modality was maintained for a stabilization period of three months before measurement of sICAM-1 levels.

Figure 1 Flow chart of study process. LFHD: Low flux hemodialysis; HFHD: High flux hemodialysis; OL-HDF: Online hemodiafiltration.

Laboratory tests

Five milliliters of venous blood were obtained from both patients and healthy controls for measurement of complete blood count, C-reactive protein (CRP), and serum sICAM-1 levels. For patients undergoing HD, samples were collected before and after a midweek dialysis session from a peripheral vein in the limb opposite to the arteriovenous fistula. Before obtaining post-dialysis samples, the blood flow rate was reduced to 50 mL/minute, while both dialysate flow and replacement fluid infusion were temporarily stopped. Samples were allowed to clot at room temperature for 10-20 minutes, whereas post-dialysis samples were left to clot for 45-60 minutes to minimize potential interference from residual heparin. The samples were subsequently centrifuged at 2000-3000 revolutions per minute for 20 minutes. Two milliliters were drawn into ethylene diamine tetraacetic acid-anticoagulated tubes for complete blood count evaluation. Hematological indices - including hemoglobin concentration, platelet count, total leukocyte count (TLC), absolute neutrophil count and absolute lymphocyte count (ALC) - were measured using an automated hematology analyzer (Coulter LH 750; Beckman Instruments Inc., Fullerton, CA, United States). For biochemical assessment, serum CRP levels were determined using a COBAS C6000 analyzer (Roche Diagnostics GmbH, Mannheim, Germany). Additionally, two milliliters of separated serum were utilized for quantification of sICAM-1 using a commercially available Human ICAM-1 enzyme-linked immunosorbent assay kit (Human ICAM-1 Enzyme-Linked Immunosorbent Assay Kit, Cat. No. E0212Hu; Bioassay Technology Laboratory, Shanghai, China). To account for hemoconcentration occurring during dialysis, post-dialysis ICAM-1 was adjusted according to the patient’s hematocrit using the van Beaumont formula, yielding corrected ICAM-1 (cICAM-1) (equation 1)[13].

Post-dialysis cICAM-1 = 1 - Δ × postdialysis ICAM-1, where Δ = postdialysis hematocrit - predialysis hematocrit/postdialysis total hematocrit (equation 1).

Dialysis prescription

Initially, all patients were maintained on conventional low-flux hemodialysis (LFHD) through arteriovenous fistula using Fresenius polysulfone dialyzers with surface areas of 0.8 m² in 16 patients (61.5%) and 1.0 m² in 10 patients (38.5%). During both high-flux hemodialysis (HFHD) and OL-HDF phases, all patients used Fresenius polysulfone dialyzers with a surface area of 0.7 m². Pediatric bloodlines were used during LFHD and HFHD, whereas hemodiafiltration-specific bloodlines were utilized during OL-HDF treatments. Unfractionated heparin was administered as the anticoagulant in all dialysis treatments. Dialysis machines included Bellco Formula, Bellco Formula Therapy, Fresenius 5008S, Gambro AK 200 ULTRA S, Gambro AK 96, and Gambro Artis systems. All dialysis sessions were performed using ultrapure water delivered through a DIASAFE^® filter and a central dialysis fluid delivery system. Dialysate flow was maintained at 500 mL/minute, with dialysate composition of sodium 138 mmol/L, chloride 106.5 mmol/L, potassium 2 mmol/L, magnesium 0.5 mmol/L, calcium 1.75 mmol/L, bicarbonate 32 mmol/L, and a minor amount of acetate to stabilize the dialysate (6 mmol/L). Appropriate bicarbonate concentrate was freshly diluted and used during dialysis sessions from Bibag Fresenius and Bicart Gambro.

Study outcomes: (1) To measure circulating cICAM-1 levels in children on HD; (2) To evaluate the effect of each dialysis modality on post-session cICAM-1; and (3) To compare the impact of different dialysis modalities on post-dialysis circulating cICAM-1 levels.

Sample size calculation: The sample size was estimated to detect a significant difference in post-dialysis sICAM-1 levels between dialysis modalities with a statistical power of 80% and a two-sided alpha level of 0.05. Assuming an effect size of (0.78-1.06), based on previously published studies[5,14,15], a minimum sample size of 22 patients per group was considered sufficient to detect clinically meaningful differences.

Statistical analysis

Data analysis was conducted using IBM SPSS Statistics version 26. The Shapiro-Wilk test was applied to evaluate the normality of continuous variables. Variables with normal distribution were expressed as mean ± SD, whereas non-normally distributed variables were presented as median [interquartile range (IQR)]. Categorical data were summarized as n (%). Comparisons between independent groups were performed using Student’s t-test for normally distributed variables and the Mann-Whitney U test for non-parametric data. Paired comparisons within the same group were analyzed using either the paired t-test or the Wilcoxon signed-rank test, depending on the distribution of the data. Correlations between variables were evaluated using Spearman’s correlation coefficient. For normally distributed variables, differences were reported as mean difference with 95% confidence intervals (CIs), and for non-normally distributed variables, the median difference was estimated using the Hodges-Lehmann median difference estimator with corresponding 95%CI. P-values were calculated for all comparisons. Because multiple pairwise comparisons were performed across dialysis modalities, Bonferroni correction was applied to reduce the risk of type I error. A P < 0.05 was considered statistically significant.

RESULTS

The HD group included 26 children (10 girls and 16 boys) with a mean age of 12.2 ± 2.8 years (range 5.8-15.6 years). The control group comprised 26 children (13 girls and 13 boys) with a mean age of 10.9 ± 2.7 years (range 6-15 years). There were no significant differences between the two groups with respect to age (P = 0.12) or sex (P = 0.80). Among HD patients, the underlying renal diagnoses were distributed as follows: Unknown etiology in 8 patients (30.8%), congenital anomalies of the kidney and urinary tract in 7 (26.9%), chronic glomerulopathy in 6 (23.1%), autosomal recessive polycystic kidney disease (ARPKD) in 2 (7.7%), Joubert syndrome in 2 (7.7%), and Alport syndrome in 1 patient (3.8%). The median HD duration was 35.5 months with an IQR of 32.6 months. Clinically, all patients were anuric, and 14 (53.85%) were hypertensive. All patients were medically stable without acute or chronic conditions that could elevate CRP, and their serum CRP levels were within the normal range. Median (IQR) CRP values were 1.3 (2.7) mg/L in LFHD, 0.9 (1.2) mg/L in HFHD, and 1.1 (1.7) mg/L in OL-HDF, showing no significant differences between dialysis modalities and comparable to controls (1.4 mg/L; 0.06 mg/L). The median blood flow rate (mL/minute/m²) was 229.7 (IQR 31.45) in LFHD, 231.7 (IQR 28.3) in HFHD, and 235.2 (IQR 24.6) in OL-HDF. Dialysis adequacy, expressed as single-pool Kt/V, showed median values of 1.30 (0.1), 1.35 (0.2), and 1.43 (0.18) for LFHD, HFHD, and OL-HDF, respectively. In the OL-HDF group, the median convection volume achieved was 13.24 L/m² (IQR 2.5).

Pre-session sICAM-1 levels in the LFHD, HFHD and OL-HDF modalities were significantly elevated compared with the control group. Pre-session sICAM-1 levels were comparable between LFHD and both of HFHD and OL-HDF, as well as between HFHD and OL-HDF. LFHD was associated with a further post-session increase in cICAM-1 which were significantly higher than both pre-session levels and controls. In contrast, HFHD resulted in a statistically significant reduction in post-session cICAM-1 compared to the pre-session levels, but remained significantly higher than the controls. OL-HDF was associated with a marked reduction of post-session cICAM-1 levels which were not only significantly lower than pre-session levels, but also comparable to those of the control group. The magnitude of post-dialysis cICAM-1 changes differed significantly across HD modalities. LFHD was associated with a median increase of 871.5 (95%CI: 373.2-1749.9; P = 0.002), whereas HFHD resulted in a reduction of -1087.1 (95%CI: -1439.5 to -755.6; P < 0.001). OL-HDF produced the greatest decrease in circulating cICAM-1 levels (-1372.9; 95%CI: -2143.8 to -786.9; P < 0.001), indicating its superior efficacy in reducing circulating cICAM-1. These findings are summarized in Tables 1 and 2, and illustrated in Figures 2 and 3. Neither dialyzer surface area in LFHD nor pump blood flow rate in all the three HD modalities were shown to significantly affect post-session cICAM-1 levels (P > 0.05 for all). Pre-session sICAM-1 concentrations were also comparable between hypertensive and non-hypertensive patients in LFHD, HFHD, and OL-HDF (P = 0.411, P = 0.382, and P = 0.264, respectively). No significant association was observed between pre-session sICAM-1 and HD duration in LFHD (P = 0.674). Similarly, pre-session sICAM-1 was not correlated with CRP levels in LFHD (P = 0.904), HFHD (P = 0.930), or OL-HDF (P = 0.621). Additionally, neither pre- nor post-session sICAM-1 showed significant associations with hemoglobin, TLC, or absolute neutrophil count and ALC (P > 0.05 for all).

Figure 2 Boxplots of post-session corrected intercellular adhesion molecule-1 vs pre-session soluble intercellular adhesion molecule-1 across different dialysis modalities. ICAM-1: Intercellular adhesion molecule-1; cICAM-1: Corrected intercellular adhesion molecule-1; sICAM-1: Soluble intercellular adhesion molecule-1; LFHD: Low flux hemodialysis; HFHD: High flux hemodialysis; OL-HDF: Online hemodiafiltration.

Figure 3 Linear graph of serum soluble intercellular adhesion molecule-1 median trend with various dialysis modalities. ¹Wilcoxon signed rank test. ICAM-1: Intercellular adhesion molecule-1; cICAM-1: Corrected intercellular adhesion molecule-1; LFHD: Low flux hemodialysis; HFHD: High flux hemodialysis; OL-HDF: Online hemodiafiltration.

Table 1 Values of serum intercellular adhesion molecule-1 levels.

Parameters	HD modality	Cases		Controls		P value	Mean difference (95%CI)
		mean ± SD	Median (IQR)	mean ± SD	Median (IQR)
Pre-session ICAM-1 (ng/L)	LFHD	1174.48 ± 1221.36	676.55 (493.88-1426.00)	174.61 ± 55.50	171.15 (125.23-226.98)	< 0.0011	523.63 (391.0-784.2)
	HFHD	1665.82 ± 1566.27	1080.00 (785.75-2009.50)			< 0.0011	929.63 (691.2-1496.4)
	OL-HDF	1493.76 ± 863.61	1207.00 (640.05-2529.00)			< 0.0011	1045.73 (593.0-1998.1)
Post-session cICAM-1 (ng/L)	LFHD	2217.18 ± 1675.90	1481.50 (1147.30-2993.10)			< 0.0011	1312.63 (1056.8-2133.2)
	HFHD	316.51 ± 262.17	240.97 (148.91-390.61)			0.0261	62.23 (6.9-144.4)
	OL-HDF	172.24 ± 74.83	163.63 (121.93-204.36)			0.8902	-2.44 (-39.3-34.6)

¹Mann-Whitney test.

²Student t test.

³Hodges-Lehman median difference estimate.

⁴Mean difference.

ICAM-1: Intercellular adhesion molecule-1; cICAM-1: Corrected intercellular adhesion molecule-1; HD: Hemodialysis; LFHD: Low flux hemodialysis; HFHD: High flux hemodialysis; OL-HDF: Online hemodiafiltration; IQR: Interquartile range; CI: Confidence interval.

Table 2 Comparisons of serum intercellular adhesion molecule-1 levels.

Cases		P value	Mean difference (95%CI)
Pre-session sICAM-1	LFHD vs HFHD	0.0581	378.92 (-10.1-864.6)
	LFHD vs OL-HDF	0.0511	737.72 (100.9-1449.0)
	HFHD vs OL-HDF	0.7511	13.32 (-364.2-812.4)
Post-session cICAM-1	LFHD vs HFHD	< 0.0011	-1655.62 (-2633.7 to -1088.0)
	LFHD vs OL-HDF	< 0.0011	-1727.22 (-2782.8 to -1188.8)
	HFHD vs OL-HDF	0.0111	-84.12 (-173.7 to -25.3)
Pre-session vs post-session cICAM-1	LFHD	0.0021	871.52 (373.2-1749.9)
	HFHD	< 0.0011	-1087.12 (-1439.5 to -755.6)
	OL-HDF	< 0.0011	-1372.92 (-2143.8 to -786.9)

¹Wilcoxon signed ranks test.

²Hodges-Lehman median difference estimate.

sICAM-1: Soluble intercellular adhesion molecule-1; cICAM-1: Corrected intercellular adhesion molecule-1; LFHD: Low flux hemodialysis; HFHD: High flux hemodialysis; OL-HDF: Online hemodiafiltration; CI: Confidence interval.

DISCUSSION

This study demonstrated significantly higher baseline circulating sICAM-1 levels in HD children compared with healthy controls[5,14], consistent with earlier reports[16-20]. These findings support the concept that children with CKD on maintenance HD exhibit a state of persistent endothelial activation and inflammation. The mechanisms contributing to elevated pre-session ICAM-1 in CKD patients are not fully understood considering divergent study results. Chronic uremia may contribute, as it is linked to increased levels of cytokines that promote inflammation, which can enhance the expression and release of various adhesion molecules[21]. In addition, high ICAM-1 levels may result from enhanced endothelial synthesis and release stimulated by pro-inflammatory cytokines triggered and synthesized during dialysis leukocyte- extracorporeal circuit interactions[3,22]. Furthermore, the observation of elevated levels of sICAM-1 in un-dialyzed CKD patients and those on chronic ambulatory peritoneal dialysis suggests a possible role of the kidneys in its catabolism[5,21]. Surprisingly, a study comparing sICAM-1 levels in children with CKD on conservative management, peritoneal dialysis, and HD reported no difference between HD patients and their control group, whereas CKD and peritoneal dialysis patients had significantly higher ICAM-1 levels[21]. However, the control group in this study was not entirely healthy, consisting of children with urinary tract abnormalities or stones with normal kidney function, conditions potentially associated with low-grade inflammation. This methodological limitation may have masked elevations in HD patients, highlighting the importance of selecting truly healthy controls and explaining why some findings appear contradictory to those observed in studies reporting higher sICAM-1 in pediatric HD patients. In line with this, Liakopoulos et al[22] similarly, reported comparable serum ICAM-1 levels between adult HD patients and healthy controls. However, differences in patient age, comorbidities, and dialysis characteristics may partly explain these discrepancies.

To our knowledge, few pediatric studies have evaluated the impact of dialysis membrane and modality on circulating sICAM-1 levels. Our results showed a significant rise in post-session corrected sICAM-1 concentrations following LFHD, whereas their levels dramatically dropped following HFHD and OL-HDF sessions. These results suggest that dialysis modality and membrane performance may influence endothelial activation during HD.

Data on the effects of different HD modalities on sICAM-1 levels in children undergoing HD are limited. In the only pediatric study conducted by Sawires et al[14], baseline ICAM-1 levels were measured in 80 children with ESKD, who were then non-randomly allocated to two equal groups receiving either LFHD or HFHD using polysulfone membranes over a three-month period. Pre-session arterial and post-session venous samples were drawn to assay ICAM-1 levels. Baseline ICAM-1 levels were comparable between the low-flux and high-flux groups, suggesting no initial group differences and minimal bias. In the LFHD group, no significant change was observed in ICAM-1 between the start and end of the study; however, the authors did not specify which comparisons were statistically tested - baseline vs either pre-session, or baseline vs post-session, or baseline vs both. Conversely, they report that high-flux HD for three months resulted in a notable reduction in pre-session ICAM-1 relative to baseline. When comparing low-flux and high-flux, both pre-dialysis, and post-dialysis ICAM-1 levels, as well as the post-dialysis increments, were significantly higher in the low-flux in comparison to the high-flux group. Clinically, this study highlights that the selection of efficient dialysis modalities using high-flux may help reduce inflammation and endothelial stress in pediatric patients.

Adult studies evaluating the effect of HD membranes and modalities on sICAM-1 have yielded inconsistent results. Papayianni et al[5] and Bieber et al[18] reported significant increases in ICAM-1 during conventional HD using cellulose or polysulfone membranes relative to pre-dialysis levels and or healthy controls. Conversely, Liakopoulos et al[22] reported no change in ICAM-1 levels during a 4-hour HD session with similar levels detected between low-flux cuprophane and high-flux polysulfone dialyzers. Contrary to our results, a multicenter crossover study by Vaslaki et al[15] stated no difference in post-dialysis ICAM-1 between OL-HDF and low-flux HD, although formal pre- and post-session comparisons were not performed.

ICAM-1, a member of the immunoglobulin superfamily, consists of 5 extracellular immunoglobulin-like domains, a single transmembrane hydrophobic segment, and a short cytoplasmic tail[23,24]. Its molecular weight ranges between 90 kDa and 110 kDa[25] largely influenced by the extent of glycosylation, which renders the intact ICAM-1 unlikely to be efficiently cleared by standard HD membranes. However, ICAM-1 expression is modulated by alternative splicing and proteolytic cleavage during inflammatory activation, generating multiple membrane-bound isoforms and soluble fragments, including smaller unglycosylated variants as small as 54-60 kDa[3,4,24]. Notably, shorter ICAM-1 isoforms may be more prone to proteolytic cleavage than the full-length molecule[4], leading to the formation of additional smaller fragments. Such molecular heterogeneity likely contributes to variability in measured circulating ICAM-1 levels and may influence its clearance characteristics during dialysis. Furthermore, variability in immunoassay sensitivity and specificity may influence the detection of circulating ICAM-1 fragments, which could contribute to differences in measured concentrations despite variations in their dialyzability. In this context, the lower post-dialysis sICAM-1 levels observed with high-flux HD and hemodiafiltration in this study may reflect the enhanced removal of middle-molecular-weight substances, including pro-inflammatory cytokines and uremic toxins in such HD modalities. Moreover, these modalities appear to induce less dialysis-related inflammation by reducing membrane-leukocyte interactions, thereby limiting ICAM-1 upregulation. This effect is particularly pronounced in OL-HDF, possibly due to its more biocompatible environment, which further supports efficient middle-molecule removal and attenuates inflammatory activation during dialysis[26,27]. Another possible explanation for the lower sICAM-1 levels in such modalities is proteins’ adsorption onto the membrane surface via electrostatic interactions. However, this mechanism alone cannot account for the persistently elevated ICAM-1 levels reported after low-flux HD using different membranes in multiple studies[5,18] suggesting that adsorption is unlikely to be a dominant mechanism.

Since hypertension exerts a shearing effect on endothelial cells that leads to the release of sICAM-1, we compared pre-dialysis sICAM-1 levels between hypertensive and non-hypertensive patients, but we did not find any significant differences between them. Similarly, no correlation was observed between HD duration and pre-dialysis sICAM-1. These findings, consistent with those reported by Papayianni et al[5] and Liakopoulos et al[22], suggest that elevated pre-session sICAM-1 levels in HD patients is likely multifactorial in origin. In contrast to earlier studies in children[14] we observed no significant correlation between either pre-session or post-session sICAM-1 levels and TLC, ANC, and ALC. This finding suggests that circulating sICAM-1 levels are more likely to reflect leukocyte activation rather than merely the number of circulating leukocytes.

The discrepant findings across studies may be linked to several factors, such as differences in patient age, cohort size, and the presence of comorbidities known to influence ICAM-1 levels. Many adult studies included patients with conditions such as diabetes mellitus, hypertension, amyloidosis, liver transplantation, failed renal grafts, or ongoing immunosuppression[18]. Technical aspects also play a role, such as sampling site, hemodilution, and variations in the sensitivity and specificity of enzyme-linked immunosorbent assay kits. For instance, blood collection directly from the arteriovenous fistula may artificially elevate ICAM-1 due to local endothelial injury; therefore, we avoided this technique. Similarly, prolonged interdialytic intervals can cause greater weight gain and hemodilution, potentially leading to spuriously low ICAM-1 levels. To minimize this bias, our laboratory assessments were performed during midweek sessions. Taken together, these methodological and clinical variations likely explain the discrepancies among studies and highlight the need for larger, well-controlled trials with longer periods of follow-up to clarify the impact of different dialyzers and HD modalities on sICAM-1 in both adult and pediatric populations.

Our study has several limitations. First, the lack of randomization and the absence of a crossover design with washout periods may have introduced potential bias. In addition, the relatively small sample size and the single-center design may limit the generalizability of our findings. Furthermore, a three-month duration may be insufficient to capture the long-term effects of these modalities on pre-dialysis sICAM-1 levels. Another limitation is that inflammatory cytokines such as interleukin-6 or tumor necrosis factor were not assayed. Additionally, assay of sICAM-1 in the dialysate was not performed, leaving the contribution of each HD modality to its clearance unclear. Lastly, linking sICAM-1 levels to long-term cardiovascular outcomes could have provided greater clinical relevance of these findings and should be explored in future studies.

CONCLUSION

The current study demonstrates that children on HD have high serum levels of ICAM-1 that were significantly increased after low-flux HD sessions. In contrast, high-flux HD and hemodiafiltration led to a significant reduction in post-session cICAM-1 levels. Notably, OL-HDF was more effective than HFHD as it resulted in a reduction of post-session cICAM-1 to levels that became comparable to that of the controls. These findings suggest that selecting more biocompatible and high-efficiency dialysis modalities in pediatric patients may help mitigate dialysis-related inflammation and endothelial activation, with potential benefits for long-term cardiovascular outcomes that warrant more evaluation in future studies.