Hematological parameter dynamics in critically ill patients treated with oXiris^®-based CRRT and CytoSorb^® hemoadsorption: a prospective single-center study

Abstract

Background

Hemoadsorption is increasingly used as adjunctive extracorporeal support in sepsis and septic shock. Comparative data on device-specific hematologic effects remain limited. CytoSorb^® and oXiris^® are among the most widely applied hemoadsorption devices, but their impact on blood element kinetics has not been fully characterized.

Methods

In a prospective single-center observational study, 66 patients with sepsis or septic shock underwent hemoadsorption combined with continuous renal replacement therapy (CRRT) using either CytoSorb^® (n = 36) or oXiris^® (n = 30). Hemoglobin (Hb), hematocrit (Hct), erythrocytes (RBC), and platelets (PLT) were measured prior to and following hemoadsorption. Descriptive analyses and statistical tests were employed to evaluate between-group differences in hematologic parameters. Clinically relevant declines defined as ≥ 15% reduction in Hb, Hct, and RBC, and ≥ 50% in PLT were treated as events in time-to-event (TTE) analysis.

Results

CytoSorb^® was associated with significant reductions in Hb, Hct, RBC, and PLT, while no statistically significant hematologic changes were observed with oXiris^®. TTE analyses showed that CytoSorb^® led to faster declines across all parameters in reaching predefined reductions. Median times to predefined thresholds ranged from 19.6 to 27.6 h with CytoSorb^®, compared with 38.9 to 42.1 h with oXiris^®, despite shorter treatment duration in the CytoSorb^® group.

Conclusion

CytoSorb^® hemoadsorption results in more pronounced and faster hematologic deterioration compared with oXiris^®, indicating meaningful device-specific differences in blood element kinetics during extracorporeal therapy. Close hematologic monitoring is particularly important when CytoSorb^® is used. Larger multicenter studies with mechanistic biomarkers are needed to validate these findings.

Introduction

Sepsis is a life-threatening syndrome characterized by organ dysfunction caused by a dysregulated host response to infection. Triggered by bacteria, viruses, fungi, or parasites, it is strongly influenced by advanced age, immune status, and pathogen virulence [1]. Despite substantial advances in intensive care medicine, sepsis and septic shock remain among the most frequent causes of morbidity and mortality in critically ill patients worldwide [2].

Management of sepsis is based on rapid identification and control of infections, hemodynamic stabilization, and comprehensive organ support. Early administration of appropriate antimicrobial therapy and prompt hemodynamic resuscitation are fundamental steps in the initial management. In patients with persistent hypotension, vasopressor support is essential to maintain adequate tissue perfusion [3, 4]. Beyond pharmacological therapy, extracorporeal organ support techniques are increasingly being used as adjunctive and supportive treatment in septic patients, including continuous renal replacement therapy (CRRT) and blood purification techniques [1, 4]. Such interventions primarily provide organ support, while their potential role in modulating the dysregulated systemic inflammatory response associated with multiple organ dysfunction remains an area of ongoing investigation. Hemoadsorption represents an emerging extracorporeal blood purification strategy designed to remove circulating inflammatory mediators, endotoxins, and other toxic molecules from the bloodstream. This technique can be integrated into CRRT circuits or used as a standalone therapy. By adsorbing circulating mediators, hemoadsorption may help modulate the dysregulated inflammatory response and potentially contribute to improved hemodynamic stability and organ function. However, because the process is largely non-selective and may remove both pro- and anti-inflammatory cytokines, its overall clinical impact remains complex and incompletely understood, and important knowledge gaps persist in the field of blood purification therapies [5, 6]. Several commercially available systems have been developed. The CytoSorb^® cartridge contains highly porous polymer beads capable of binding hydrophobic molecules, such as cytokines and toxins. It is designed to reduce the excessive cytokine load observed in hyperinflammatory states such as sepsis. The oXiris^® hemofilter, on the other hand, combines CRRT with adsorption of endotoxins and cytokines, enabling simultaneous detoxification and renal support [7].

While definitive clinical benefits of hemoadsorption have yet to be established, its use as an adjunctive modality is growing among septic patients undergoing extracorporeal support [8–11]. Understanding their potential impact on hematological parameters is important, as these markers primarily reflect potential therapy-related effects on blood components. However, they may also be influenced by the underlying severity of sepsis and sepsis-associated coagulopathy, making it difficult to distinguish device-related effects from disease-related changes [12–14]. Accordingly, this study aims to assess the influence of CytoSorb^® and oXiris^® hemoadsorption on hematological parameters in critically ill patients with sepsis and septic shock.

Methods

Ethical aspects

This prospective observational study was conducted in the 28-bed Medical Intensive Care Unit (MICU) at the University Clinical Centre of the Republic of Srpska, a tertiary-level, university-affiliated hospital, from September 1, 2023, to December 31, 2024. The protocol was approved by the Ethics Committee of the University Clinical Centre of the Republic of Srpska (approval no. 01-19-373-2/23) and the research was conducted in accordance with the Declaration of Helsinki. In patients who were unable to provide informed consent due to impaired consciousness or cognitive decline, informed consent was obtained from their legally authorized representatives. Trial registration: ClinicalTrials.gov (https://clinicaltrials.gov/): NCT06261164 (Feb 1, 2024).

Patients

All adult patients (≥ 18 years) admitted with sepsis or septic shock according to Sepsis-3 criteria [15], who had sepsis-associated acute kidney injury (SA-AKI) requiring renal replacement therapy and were treated with a hemoadsorption device (CytoSorb^® or oXiris^®), were eligible for inclusion. Hemoadsorption therapy was initiated as an adjunctive treatment in patients with severe systemic inflammation and evolving organ dysfunction, based on the clinical judgment of the treating intensivist, and was not restricted exclusively to patients requiring vasopressor support. This was a prospective, single-center, observational study conducted over a predefined 18-month period. During this time frame, all consecutive patients who met the predefined inclusion and exclusion criteria were screened and enrolled. The study duration was determined a priori based on feasibility considerations within routine clinical practice in a low-resource setting, including the anticipated availability of hemoadsorption devices. Patients were excluded if the primary diagnosis was not septic shock, such as out-of-hospital cardiac arrest, rhabdomyolysis, intoxications, metabolic disturbances, kidney or heart failure with type 1 respiratory failure requiring CRRT, or active malignancy, as those might confound the procedures, including CYTOSORB therapy, Molecular Adsorbent Recirculation System (MARS, BAXTER, Deerfield, IL, USA), plasmapheresis, CO₂ elimination using PRISMAX (BAXTER, Deerfield, IL, USA), a history of heparin-induced thrombocytopenia (HIT), or heparin allergy. Management of sepsis and septic shock utilizing antimicrobial therapy, fluid resuscitation, vasopressor support, and renal replacement therapy was performed following the institutional protocol and current international guidelines (Surviving Sepsis Campaign 2021) [16].

Continuous renal replacement therapy (CRRT) and hemoadsorption

For all patients with AKI and sepsis and septic shock receiving CRRT, the prescribed effluent dose was set at 25–30 mL/kg/h to account for circuit downtime and treatment interruptions, to achieve a delivered effluent dose of approximately 20–25 mL/kg/h, consistent with guideline-based targets. CRRT was performed using the PRISMAFLEX system (Baxter International Inc., USA), which supports multiple modalities, including continuous venovenous hemodiafiltration (CVVHDF), continuous venovenous hemodialysis (CVVHD), continuous venovenous hemofiltration (CVVH), slow continuous ultrafiltration (SCUF), and therapeutic plasma exchange (TPE). CVVHDF as a mode of CRRT was used for the treatment of all patients included in the study. Ultrafiltration was applied in both treatment groups as part of routine CRRT management, with fluid removal rates not exceeding 2 mL/kg/h of ideal body weight, consistent with conservative fluid removal strategies recommended for hemodynamically unstable patients with sepsis-associated acute kidney injury. The choice of anticoagulation strategy was based on patients’ clinical parameters (e.g., platelet count, INR) and comprised either citrate or heparin anticoagulation, as prescribed. Hemoadsorption therapy was initiated only after completion of the initial resuscitation phase according to Surviving Sepsis Campaign recommendations, including fluid resuscitation, initiation of vasopressor support when required, and initial hemodynamic stabilization. The timing of hemoadsorption initiation was based on clinical judgment after initial stabilization and was not determined by device availability. The CytoSorb^® cartridge (CytoSorbents Corporation, USA) is a sterile, non-pyrogenic, single-use hemoadsorption device containing highly porous divinylbenzene copolymer beads designed to remove a broad range of hydrophobic inflammatory mediators and cytokines up to approximately 55 kDa through size exclusion and hydrophobic interactions. The cartridge was integrated in series into the CRRT circuit, positioned after the hemofilter in the post-dialyzer (venous) line, according to the manufacturer’s recommendations [10]. Each CytoSorb^® cartridge was used for 12–24 h in accordance with the manufacturer’s recommendations, after which the cartridge was removed from the circuit. The oXiris^® hemofilter (Baxter International Inc., USA) utilizes an AN69-based membrane composed of acrylonitrile and sodium methallyl sulfonate copolymer fibers, surface-treated with polyethyleneimine and pre-grafted with heparin (approximately 4500 ± 1500 IU/m²). This system enables simultaneous renal replacement therapy, cytokine and endotoxin adsorption, and antithrombogenic surface protection. The membrane has an effective surface area of 1.5 m², an internal fiber diameter of 240 μm, and a wall thickness of 50 μm. The duration of each hemoadsorption session was recorded at the end of treatment for every patient [17, 18]. The choice of hemoadsorption device (CytoSorb^® or oXiris^®) was made by the treating physician as part of standard clinical care. oXiris^®-based CRRT was preferentially selected in patients with suspected Gram-negative sepsis, most often of nosocomial origin, based on clinical findings and available microbiological data, given the suspected presence of endotoxemia. Direct endotoxin measurements were not available in our laboratory and, therefore, could not be used to guide device selection. All patients received empiric antibiotic therapy in accordance with local microbiological resistance patterns, and microbiological samples were obtained before the initiation of empiric antibiotics whenever feasible.

Data collection and laboratory analysis

According to the study protocol, blood samples were collected 15–30 min before initiation of hemoadsorption with either CytoSorb^® or oXiris^® and immediately upon completion of the procedure. Hematological analyses included hemoglobin (Hg), hematocrit (HCT), erythrocyte count (RBC), platelet count (PLT), and leukocyte count. Comprehensive demographic and clinical data were collected both at the time of MICU admission and at the initiation of hemoadsorption. The collected variables included age, sex, body height, body weight, vasopressor use, duration of hemoadsorption, length of hospital stay, and therapeutic outcome (survival). The Simplified Acute Physiology Score II (SAPS II) was assessed upon MICU admission, while the Sequential Organ Failure Assessment (SOFA) score was recorded on the day of biochemical and hematological sampling. All hematological tests were performed in the hospital’s Department of Laboratory Diagnostics using the Sysmex XN-3100™ hematology analyzer, which employs fluorescence flow cytometry (WBC DIFF/RET/PLT-F/WPC) with an integrated SP-50 module for automatic smear preparation and staining. Biochemical analyses were carried out using the Abbott Alinity C analyzer (Abbott Diagnostics, USA), which utilizes photometric and potentiometric detection methods.

Statistical analysis

Statistical analyses were performed using SPSS Statistics for Windows, version 30.0 (IBM Corp., Armonk, NY, USA). Continuous variables are expressed as medians with interquartile ranges (IQRs), whereas categorical variables are presented as absolute counts and percentages. The Kolmogorov–Smirnov and Shapiro–Wilk tests were applied to assess the normality of data distribution. Differences in continuous variables between groups were analyzed using the Mann–Whitney U test for independent samples and the Wilcoxon signed-rank test for paired comparisons. Categorical variables were compared using the chi-square (χ²) test, Fisher’s exact test, or Pearson’s chi-square test, as appropriate. A two-tailed p-value < 0.05 was considered statistically significant.

Time-to-event analysis

Based on group-wise comparisons of parameter reductions during hemoadsorption, new categorical variables were created to reflect clinically relevant decreases in Hb, HCT, RBC, and PLT in patients treated with CytoSorb^® or oXiris^®. Reduction in Hb, Hct, or RBC of at least 15%, or a at least 50% in PLT following hemoadsorption compared to its initiation was considered as clinically relevant. Hence, an event indicated a decrease meeting or exceeding these thresholds, while no-event indicated a smaller or absent decrease in hematological parameters during hemoadsorption. We applied both non-parametric and parametric time-to-event (TTE) approach to account for differences in hemoadsorption duration across patients, and to evaluate the impact of hemoadsorbent type together with other clinically relevant covariates. Variables showing the significance in univariate analyses were included in multivariate parametric survival models. Final parametric models were chosen based on clinical plausibility of variables, collinearities among variables, and improvement of model fit based on Akaike Information Criterion (AIC) value. During parametric approach, comparisons across candidate hazard distributions (Weibull, exponential, log-normal, log-logistic, and Gompertz) were performed, and the model with the lowest AIC was chosen. In parametric TTE models, instantaneous event risk is described through the hazard function (h(t)), and survival probability is derived directly from the cumulative hazard (H(t)) over time using the relationship: , representing the probability of remaining event-free at each time point. Median survival times were calculated from the fitted models for each combination of significant covariates to quantify the expected time to a predefined decrease in biochemical parameter. Statistical significance was defined as p<0.05. All analyses were performed in R (version 4.5.1), RStudio 2025.09.0, using the survival and flexsurv packages.

Results

In this clinical study, a total of 69 patients were enrolled. Serum samples were obtained from all participants both prior to and following hemoadsorption (n = 69 pre-treatment; n = 69 post-treatment samples). Among the post-hemoadsorption samples, three samples were deemed inadequate for hematological analysis. Demographic and clinical characteristics of the study population are summarized in Table 1. Median (IQR) SAPS II score at admission was 49 (37) in the Cytosorb^® group and 59 (19) in the oXiris^® group. The median (IQR) SOFA scores for patients treated with Cytosorb^® and oXiris^® were 12 (5) and 11 (7), respectively. The median (IQR) length of hospital stay was 9 (22) days for Cytosorb^® patients and 16 (16.5) days for those treated with oXiris^®. Overall, we observed homogeneity between the Cytosorb^® and oXiris^® patient groups. A statistically significant difference was found in the median (IQR) duration of hemofiltration between the two groups: 15 (8) hours in the Cytosorb^® group and 24 (16) hours in the oXiris^® group (p = 0.005). No significant differences were observed between the Cytosorb^® and oXiris^® groups regarding lactate kinetics or hemodynamic response. Absolute lactate reduction and lactate clearance were comparable between groups (p = 0.63 and p = 0.14, respectively), as were absolute and relative changes in MAP (p = 0.69 and p = 0.94). Similarly, no differences were found in ICU length of stay, duration of vasopressor therapy, or ICU mortality (all p > 0.05).

Table 1.

Demographic and clinical characteristics of patients treated with CytoSorb^® and oXiris^®

Characteristic (units)	Cytosorb®	oXiris®	p value
Sex (male); N (%)	27 (54)	23 (46)	0.994
Aged over 65 (years); N (%)	12 (52.17)	11 (47.83)	0.732
Survivors; N (%)	19 (65.52)	10 (34.48)	0.168
Lactate (mmol/L); median (IQR)	2.20 (2.42)	2.1 (1.80)	0,995
Vasoppresor; N (%)	26 (52)	24 (48)	0.405
Age (years); median (IQR)	54 (19)	57 (52)	0.491
Weight (kg); median (IQR)	89 (41)	89 (7)	0.941
Height (cm); median (IQR)	183 (14)	182 (14)	0.323
SAPS II; median (IQR)	49 (37)	59 (19)	0.351
SOFA; median (IQR)	12 (5)	11 (7)	0.451
Blood flow, (mL/min), mediana (IQR) Dialysate flow (mL/h), mediana (IQR) Dialyse typ (mL/h), heparin, N (%) LH (days); median (IQR)	150 (43) 1200 (200) 19 (47.5) 9 (22)	150 (50) 1200 (200) 21 (52.5) 16 (16.50)	0.204 0.256 0.138 0.257
LF (hours); median (IQR)	15 (8)	24 (16)	0.005
Absolute changes in lactate levels, (mmol/L)	-1.2 [-2.8- -0.5]	-1.7 [-2.8- -0.5]	0.63
Lactate clearance, (%)	-39.7 [-47.2- -19.3]	-44.5 [-54.0- -33.3]	0.14
Absolute changes in MAP, mmHg	5 [5–10]	5 [5–11]	0.69
Relative changes in MAP, (%)	9.6 [7.7–18.2]	9.8 [7.7–16.7]	0.94
ICU length of stay, day	11 [3–24]	10 [4-122]	0.81
Vasopressor duration, day	4.0 [2.0–8.0]	4.0 [2.0–6.0]	0.72
ICU mortality, n (%)	20 (53)	15 (47)	0.52

N, number; SAPS II, simplified acute physiology score; SOFA, Sequential Organ Failure Assessment; LH, Length of hospitalization; LF, Length of filtration; IQR, interquartile range; ICU, intensive care unit; MAP, mean arterial pressure; HR, heart rate. Data are expressed as mean ± SD or as median (interquartile range, 25–75) or as count and percentage. Absolute changes were calculated as the post-hemoadsorption value minus the pre-hemoadsorption value. Relative changes were calculated as the absolute change divided by the pre-hemoadsorption value, multiplied by 100

Before initiation of hemofilters, Hb, HCT, RBC, and PLT values generally were not within the reference range, consistent with the severity and complexity of critical illness. Individual effects of Cytosorb^® and oXiris^® hemoadsorption on changes, specifically reductions, in hematological parameters (Hb, HCT, RBC, and PLT) are reported in Tables 2 and 3. A statistically significant decrease in hematological parameters was observed following Cytosorb^® hemoadsorption for Hb (p = 0.02), HCT (p = 0.005), RBC (p = 0.024), and PLT (p < 0.001). In contrast, hemoadsorption with oXiris^® did not result in statistically significant changes in hematological parameters.

Table 2.

Hematological parameters before and after hemoadsorption with CytoSorb^®

Characteristic (units)	Before CytoSorb®	After CytoSorb®	p value
	Median (IQR)
Hb (g/L)	104 (32.25)	97.50 (21.25)	0.02
HCT (L/L)	0.32 (0.09)	0.29 (0.06)	0.005
RBC (1012/L)	3.46 (1.05)	3.24 (0.79)	0.024
WBC [109/L)	9.82 (17.3)	9.85 (13.63)	0.241
PLT (109/L)	113.50 (133)	73 (92.75)	< 0.001

IQR, interquartile range; Hb, hemoglobin; HCT, hematocrit; RBC,red blood cells; WBC, white blood cells; PLT, platelets

Table 3.

Hematological parameters before and after hemoadsorption with oXiris^®

Characteristic (units)	Before oXiris®	After oXiris®	p value
	Median (IQR)
Hb (g/L)	99 (30.50)	91 (29)	0.269
HCT (L/L)	0.30 (0.10)	0.28 (0.09)	0.117
RBC (1012/L)	3.57 (1.04)	2.92 (1.11)	0.139
WBC (109/L)	13.38 (20.99)	17.99 (21.64)	0.968
PLT (109/L)	146 (120.50)	105 (125.50)	0.104

IQR, interquartile range; Hb, hemoglobin; HCT, hematocrit; RBC, red blood cells; WBC, white blood cells; PLT, platelets

Varying degrees of decline (Tables 2 and 3) across patients treated with Cytosorb^® (n = 36) or oXiris^® (n = 30) led to the development of categorical variables to capture clinically meaningful decreases over time. The proportion of patients reaching pre-defined thresholds was higher for Cytosorb^® than oXiris^® group for HCT, RBC, and PLT, while a decrease in Hb for 15% were comparable between the two filters. However, a 10% Hb decrease was different between the two filters. In addition to filter type (Cytosorb^® or oXiris^®), vasopressor use showed a significant effect in univariate testing for Hb and HCT events (p = 0.034), but in multivariate together with filter type, its effect did not reach significance. Hence, hemofilter remains the main effect, while vasopressor use remained in the model to adjust it, as a clinically relevant covariate. Hemadsorption type (Cytosorb^® or oXiris^®) was associated with the event in univariate nonparametric analyses as shown at Fig. 1.

Fig. 1.

Kaplan–Meier survival curves stratified by type of filter (Cytosorb^® vs. oXiris^®) for events defined as remaining: panel A – hemoglobin (Hb) within 10%; panel B – hematocrit (HCT) within 15%; panel C – red blood cells (RBC) within 15%; panel D – platelets (PLT) within 50% of pre-filter value

We evaluated the effect of hemoadsorbent type on the probability of maintaining hematologic parameters within clinically relevant thresholds during hemoadsorption. Parametric survival curves were fitted using a best-fitted model (log-normal or Gompertz, Table 4), adjusting for clinically relevant covariates. Figure 2 shows the predicted probability of hematologic stability for Cytosorb^® and oXiris^®. Cytosorb^® and oXiris^® displayed distinct profiles, with median times to parameter decrease (50% probability) illustrated by the dashed line. Vasopressor use was held constant in the prediction. These curves allow visualization of the differential impact of the two hemoadsorbents on Hb, HCT, RBC, and PLT over the course of therapy. Predicted median survival times for Cytosorb^® and oXiris^® were generated using fitted model-based quantile estimates for each of the events previously defined, and adjusted by vasopressor use (Table 4).

Table 4.

Median survival times (hours) for Cytosorb^® and oXiris^® using the best-fitted time-to-event model adjusted for vasopressor

Event (reduction from initiation of hemoadsorbens)	Hb 10%	HCT 15%	RBC 15%	PLT 50%
Model for hazard	lnorm	Gompertz	Gompertz	lnorm
Cytosorb®	24.78	27.61	27.61	19.62
oXiris®	38.89	41.30	41.27	42.11

Hb, hemoglobin; HCT, hematocrit; RBC,red blood cells; WBC, white blood cells; PLT, platelets

Fig. 2.

Predicted probability of maintaining panel A – hemoglobin (Hb) within 10%; panel B – hematocrit (HCT) within 15%; panel C – red blood cells (RBC) within 15%; panel D – platelets (PLT) within 50% of pre-filter value based on time-to-event models adjusted for vasopressor use during hemoadsorption by Cytosorb^® and oXiris^®

Overall, the analysis indicates that patients treated with Cytosorb^® exhibited a fastest decline across all events evaluated. Patients treated with oXiris^® tend to remain longer time period on hemoadsorption therapy compared to those on Cytosorb^® to reach the same drop in parameters from baseline. Parametric TTE models allowed quantification of median times to clinically relevant decreases, providing actionable insight into the hematologic effects of hemoadsorption in critically ill patients.

Discussion

To the best of our knowledge, this is the first study to directly compare the hematologic effects of CytoSorb^® and oXiris^® hemoadsorption in critically ill patients with sepsis and septic shock. This study was observational, not interventional, due to no treatment protocols, procedures, or patient management strategies. The research analyzes hematological dynamics over time within a real-world clinical setting and does not investigate efficacy or safety in a controlled trial framework. Our findings show clear and clinically meaningful differences between the two devices. CytoSorb^® was consistently associated with significant reductions in Hb, HCT, RBC, and PLT, whereas oXiris^® did not produce statistically significant changes in any hematologic parameter. TTE analyses further strengthened these observations, showing that patients treated with CytoSorb^® reached predefined thresholds of hematologic decline substantially faster, approximately 50% earlier than those treated with oXiris^®. oXiris^® preserved hematologic stability markedly longer, with median times to clinically relevant decreases extending from 15 to over 20 h beyond those seen with CytoSorb^®. Our findings suggest that intrinsic device-specific characteristics, rather than patient-related factors, drive the magnitude and timing of hematologic deterioration during hemoadsorption therapy. Notably, these findings are further reinforced by the observation that CytoSorb^® sessions were significantly shorter in duration than those performed with oXiris^®, yet despite this shorter exposure time, CytoSorb^® was still associated with a markedly faster and more pronounced decline in hematologic parameters.

The hematologic changes observed in the CytoSorb^® group should also be interpreted in the context of concurrent CRRT, as all patients treated with CytoSorb^® underwent continuous renal replacement therapy. CRRT alone is known to contribute to declines in hematologic parameters through extracorporeal circuit–mediated platelet activation, complement activation, and contact-driven consumption, resulting in adhesion or sequestration of platelets within the hemofilter [19, 20]. Mechanical shear forces across the filter membrane, hemolysis due to circuit turbulence, and transmembrane losses may further reduce circulating erythrocytes and hemoglobin [21]. The addition of a CytoSorb^® cartridge extends the extracorporeal surface area and blood–material interface, which may potentiate these effects.

Prior reports have described measurable decreases in PLT during CytoSorb^® therapy, potentially due to adsorption, activation, or retention of platelets within the polymer matrix [22]. Although direct clinical evidence is limited, it is plausible that similar mechanisms, including blood–material contact activation, shear-stress–induced microhemolysis, and turbulence within the extracorporeal circuit, may contribute to declines in Hb, RBC, and HCT. However, published data specifically addressing these erythroid changes during CytoSorb^® therapy remain scarce. Thus, the combination of CRRT and hemoadsorption likely produces a cumulative effect on blood element loss, providing a physiologic explanation for the more rapid and pronounced hematologic deterioration observed in patients treated with CytoSorb^® compared with oXiris^®, despite shorter treatment duration. Although the decrease in PLT observed in the oXiris^® group was not statistically significant in our cohort, similar findings have been reported by other investigators [23], and the overall body of literature specifically evaluating hematologic effects of oXiris^® therapy remains limited. Nevertheless, several mechanistic considerations warrant attention. The oXiris^® membrane is pre-grafted with heparin, a feature designed to enhance hemocompatibility, yet heparin exposure can potentially induce thrombocytopenia and increase the risk of hemorrhage [23]. In addition, experimental data suggest that portions of the extracorporeal circuit not coated with heparin may activate coagulation pathways, consume coagulation factors, and contribute to platelet decline [24]. Taken together, these considerations indicate that while clinically significant thrombocytopenia is not consistently observed with oXiris^®, a modest reduction in PLT is biologically plausible and should be monitored. Importantly, although statistically significant reductions in several hematologic parameters were observed, these changes did not reach thresholds generally considered clinically critical in routine intensive care practice. In particular, platelet counts in our cohort remained well above the level considered a contraindication for CytoSorb^® therapy (< 20 G/L), suggesting that these findings primarily support careful monitoring of hematological parameters rather than limiting the use of hemoadsorption therapy. Importantly, the additional patient-centered outcomes included in our analysis did not demonstrate significant differences between the CytoSorb^® and oXiris^® groups in terms of lactate clearance, hemodynamic improvement (MAP changes), vasopressor duration, ICU length of stay, or ICU mortality. These findings suggest that while hematologic changes differed between devices, the short-term clinical course and overall patient outcomes were broadly comparable in this cohort.

This study has several strengths, including a direct comparison of two hemoadsorption devices, CytoSorb^® and oXiris^®, within the same critically ill septic population, which is rarely reported in the literature. Second, the combined use of conventional statistical tests for group comparisons and TTE analysis provides additional depth by capturing both the magnitude and time course of hematologic changes during hemoadsorption. Conducting the study in a high-volume tertiary MICU with standardized sepsis protocols further strengthens internal consistency. However, we acknowledge some limitations of this study. First, this was a single-center, non-randomized study, which introduces potential selection bias and limits generalizability. Second, given the limited time available for study completion, taking into account feasibility in routine clinical practice, and the relatively small sample size, particularly in the oXiris^® subgroup, this reduces statistical power to detect between-group differences. Additionally, baseline hematological parameters were not formally compared between groups, as the analysis primarily focused on within-group changes following hemoadsorption therapy; therefore, potential baseline differences may represent a source of residual confounding.

Furthermore, the type of anticoagulation used during CRRT (citrate or heparin) was not analyzed in relation to hematological changes and may therefore represent another potential confounding factor. Moreover, differences in treatment duration between devices, although reflective of real-world practice, may have influenced hematologic trajectories. Finally, the absence of mechanistic biomarkers, such as endotoxin levels, markers of hemolysis, or platelet activation indices, limits the ability to directly link observed hematologic changes to specific device-related mechanisms.

Conclusion

In this prospective single-center study, CytoSorb^® and oXiris^® hemoadsorption were compared with a focus on hematologic effects in septic and septic shock patients receiving CRRT. CytoSorb^® was associated with significant and faster declines in Hb, HCT, RBC, and PLT, whereas oXiris^® did not produce statistically significant hematologic changes. These findings indicate meaningful device-specific differences in the kinetics of blood element loss during extracorporeal therapy. Careful hematologic monitoring is therefore warranted, particularly when CytoSorb^® is used. Larger multicenter studies are needed to confirm these results and to clarify the underlying mechanisms and clinical implications.

Abbreviations

CRRT

Continuous renal replacement therapy

Hb

Hemoglobin

Hct

Hematocrit

RBC

Erythrocytes

PLT

Platelets

TTE

Time-to-event

HIT

Heparin-induced thrombocytopenia

CVVHDF

Continuous venovenous hemodiafiltration

CVVHD

Continuous venovenous hemodialysis

CVVH

Continuous venovenous hemofiltration

SCUF

Slow continuous ultrafiltration

TPE

Therapeutic plasma exchange

MICU

Medical Intensive Care Unit

SAPS II

The Simplified Acute Physiology Score II

SOFA

Sequential Organ Failure Assessment

IQR

Interquartile ranges

AIC

Akaike Information

LH

Length of hospitalization

LF

Length of filtration

N

Number

Author contributions

NŠ: Conceptualization, Data curation, Formal analysis, Writing – original draft, Methodology. KV: Writing – original draft, Formal analysis, Methodology Validation. JV: Data curation. TK: Data curation, Formal analysis, SD: Data curation, BZ: Data curation. DM: Data curation. MJ: Data curation, GK: Data curation. PK; Writing – review & editing, Methodology, Conceptualization, Supervision.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

The datasets used and analyzed during the current study are not publicly available owing to ongoing studies, but are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The protocol was approved by the Ethics Committee of the University Clinical Centre of the Republic of Srpska (approval No. 01-19-373-2/23) and conducted in accordance with the Declaration of Helsinki. In patients who were unable to provide informed consent due to impaired consciousness or cognitive decline, informed consent was obtained from their legally authorized representatives.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.