Commentary: refining patient stratification for haemoadsorption in cardiac surgery: comparative reflections on RECCAS, REMOVE, and SIRAKI02

Clinical Registration: The RECCAS trial was prospectively registered (Clinical Trial Number DRKS00007928, https://drks.de/search/en/trial/DRKS00007928 on 3rd August 2015 with the Clinical Trial Registry and published under: Baumann A, Buchwald D, Annecke T, Hellmich M, Zahn PK, Hohn A. RECCAS - REmoval of Cytokines during Cardiac Surgery: study protocol for a randomised controlled trial. Trials. 2016;17: 137.

Haemoadsorption (HA) to modulate inflammation after cardiac surgery via extracorporeal cytokine removal has theoretical benefits, with divergent study results and clinical outcomes. These findings fuel debate on HA’s clinical relevance in cardiac surgery [1–3] during cardiopulmonary bypass (CPB). This commentary reflects on the findings of the RECCAS trial, a prospective, randomised trial investigating intraoperative haemoadsorption (HA) in elective cardiac surgery [4]. While HA lowered certain cytokines during surgery and improved cardiac index, fluid, and fibrinogen needs, it did not affect IL-6 at ICU nor organ function. We contextualise RECCAS findings by comparing with the REMOVE and SIRAKI02 trials [5, 6], and derive implications for patient stratification and future trial design. Among available trials, REMOVE and SIRAKI02 provide recent and relevant trials due to their prospective, randomised design and focus on HA in cardiac surgery. Despite differences in patient populations and devices, these trials allow a comparative evaluation of methodologies and directionality of outcomes.

Study population and procedural Bias mitigation

Notably, 30.5% of cardiac procedures are performed in patients aged 70–79 years [5]. The RECCAS trial included patients aged ≥ 65 years reflecting the typical demographic undergoing elective cardiac surgery. While this focus enhances relevance for the typical older cardiac surgery population, it may limit applicability to less frequently encountered groups such as younger patients. Broader inclusion criteria could enhance generalisability but may introduce heterogeneity. Although older patients show a less pronounced inflammatory response, it still contributes to complications.

In contrast, the REMOVE trial enrolled patients with infective endocarditis and a high inflammatory burden, whereas SIRAKI02 targeted on individuals with lower preoperative risk profiles. Future studies should investigate age-related differences across broader populations and focus on subgroup analyses, particularly in patients with pronounced inflammatory activity.

The RECCAS cohort included isolated coronary artery bypass grafting (CABG) and valve surgeries but also complex combined procedures (Supplement 1), reflecting the diversity of real-world surgical practices [5–7]. Although full blinding was not possible, restricted visibility for surgeons and concealed group allocation reduced performance bias. Comparable intraoperative times and full ICU blinding confirm objective outcome assessment.

Inflammatory monitoring and outcome assessment

RECCAS focused on IL-6 as primary marker of inflammation, acknowledging its limitations due to kinetic variability. The HA efficacy depends on CPB duration and cytokine gradient [8, 9], but the relatively low inflammatory burden, CPB-restricted HA application and potentially early adsorber saturation may have limited impact on IL-6. CPB-related cytokine release may persist postoperatively, potentially limiting the impact of intraoperative HA [10]. Neither were differences in CRP, PCT, creatinine, bilirubin, blood gas analyses, leucocytes, thrombocytes, glomerular filtration rate (eGFR), blood urea nitrogen (BUN), liver enzymes, fibrinogen, coagulation markers, SOFA-scores and ΔSOFA-scores detected. Those parameters were less suitable as primary outcome due to CRP’s delayed kinetics, creatinine’s variation based on muscle mass, and bilirubin’s sensitivity to haemolysis. Similarly CRP and procalcitonin (PCT) did not show significant differences in the REMOVE-trial [11]. The statistical robustness of RECCAS was ensured through adherence to pre-specified comprehensive protocols and transparent reporting. Accordingly, one patient was excluded post-randomisation due to a protocol violation unrelated to HA intervention [4], without compromising the intention-to-treat (ITT) principle. Sensitivity analyses confirmed the reliability of the findings.

To ensure a homogeneous cohort, immunosuppressed patients were excluded based on predefined criteria. Glucocorticoids were not routinely administered, and transfusion practices were comparable between groups, minimizing procedural or treatment-related confounders. No increased need for transfusion, coagulation factors or transfusion-associated inflammatory burden, was observed intraoperatively or during ICU (Supplement 2).

Notably, in the SIRAKI02-trial particularly high-risk patients with comorbidities, chronic kidney disease or reduced cardiac function benefitted of HA (Oxiris connected to CKRT), suggesting a relevance for cytokine burden and disease severity [11, 12], in contrast to RECCAS.

REMOVE, which included patients with a higher inflammatory baseline, similarly found no significant effect on SOFA trajectories. HA efficacy may depend on identifying patients with substantial inflammatory activity and prolonged CPB times.

Comparative insights and methodological reflections of RECCAS, REMOVE, and SIRAKI02

Key differences among RECCAS, REMOVE, and SIRAKI02 reflect the broad heterogeneity in HA research and investigated populations (Table 1). The REMOVE study included emergency and urgent endocarditis patients with elevated EURO and SOFA scores. In contrast, the RECCAS study focused on elective cardiac surgery without preoperative inflammatory processes. The SIRAKI02 study enrolled non-emergent cardiac surgery patients with even lower EURO (~ 2.5%) and SOFA scores (~ 6). Furthermore, the interventions differed: Cytosorb was used in REMOVE and RECCAS, and Oxiris in SIRAKI02. The trials also varied in their primary endpoints, which influenced power calculations and statistical methodology, underlining heterogeneity and the problematic direct comparison. REMOVE focused on Delta-SOFA within 9 days, RECCAS examined IL-6 levels at ICU admission, and SIRAKI02 assessed the incidence of CSA-AKI by day 7. The inflammatory response following cardiac surgery is multifactorial, involving ischaemia-reperfusion, inflammation, oxidative stress, haemolysis, and nephrotoxins. While endotoxin release is a likely trigger for inflammation, it constitutes only one aspect of a complex pathophysiological process. Therefore, drawing the conclusion that the trials differences in renal outcomes are primarily explained by the elimination of endotoxins may be speculative [1]. These differences preclude direct comparison but allow hypothesis generation regarding patient selection and outcome sensitivity - investigating the immune response and the influence of various mediators on outcomes and organ failure may be a valuable approach.

Table 1.

Comparative Overview of REMOVE, RECCAS, and SIRAKI02 trials

Study	REMOVE	RECCAS	SIRAKI02
Aim/Objective	Powered for superiority	Powered for significant differences	Incidence
Patient Population	Endocarditis patients	Elective cardiac surgery patients	Cardiac surgery patients with high risk for CSA-AKI
Age	68.75#	74.6 ± 5.5	69 ± 9
EURO Score	19.65%#	9 ± 28*	2,4#
Female gender	26%	37%	35%
Timing of Surgery	71%# urgent/emergency	Only elective	Nonemergent
Intervention	HA during CPB	HA during CPB	HA during CPB
HA	Cytosorb	Cytosorb	Oxiris
Integration of HA	With blood pump via the heart-lung machine (HLM) during CPB	With blood pump via the heart-lung machine (HLM) during CPB	With continuous kidney replacement therapy (CKRT)
Control	Standard of care	Standard of care	Standard of care
Primary Endpoint	Delta-SOFA within 9 days	IL-6 levels at ICU admission	Incidence of CSA-AKI by d7
Result	ΔSOFA did not differ between groups	Not difference of IL-6 at ICU admission	Significant reduction of CSA-AKI in the first 7 days after surgery
Cytokines	IL-1β, IL-18, and IL-6 significantly reduced in HA; no difference in IL-10 and TNF-α (N = 50).	IL-2, IL-6, IL-8, IL-10 reduced during surgery; no reduction at ICU admission or up to day 2 in HA.	During surgery TNF-α significantly different, no significant difference for IFN-γ, IL-2, IL-6, IL-8, IL-10
SOFA Score at ICU Admission	9#	7,8 ± 1,9 (7–9)	6
Clinical and other outcomes	No differences in mortality at 30 days, lab parameters, or secondary outcomes.	No significant differences in SOFA scores, lactate, CRP, or PCT or other laboratory parameters. Improved cardiac index. No differences in RRT need, shorter duration.	Most secondary and exploratory endpoints showed no significant differences. EBP was more effective in CSA-AKI reduction for high-risk subgroups (e.g., CKD, diabetes, low LVEF).

Comparison of key characteristics and differences across the randomised controlled trials REMOVE, RECCAS, and SIRAKI02 studies [1, 4, 5]. Differences in patient characteristics, procedural details, and outcomes are highlighted. * indicates that Euro Score I, was used instead of II to assess the risk. # - values derived or extracted from figures or text in the original publication

The future of HA in cardiac surgery

Moving forward, patient stratification and focus on complex surgical procedures, significant comorbidities, including pre-existing organ dysfunction, and clinical relevant scores (e.g. SOFA) may be essential in well-designed large-scale randomized controlled trials. Rather than enrolling heterogeneous groups, stratified trial designs could target subgroups more likely to benefit from HA, such as those with prolonged CPB, organ dysfunction, or sepsis-like profiles. Phenotype-based exploratory approaches including preoperative inflammatory phenotyping, cytokine quantification, complement activation, and markers of endothelial dysfunction should elucidate the underlying mechanisms. Standardised core outcome sets, integration of transcriptomic and proteomic analyses and early identification algorithms could enhance results.

Lessons from sepsis research suggest that combining HA with phenotype-based stratification and validated clinical scoring tools may help optimise both the timing and duration of therapy [13–15]. As we progress, the broader adoption of HA will necessitate well-powered studies that build on the discussed research. Refinement of trial methodologies, expansion of biomarker analysis, and adoption of patient-centred strategies — including optimisation of timing, dosage, duration, and patient selection algorithms—will be crucial to realise the full clinical potential of HA.

Conclusion

RECCAS, REMOVE, and SIRAKI02 collectively highlight the complexity of translating HA’s theoretical benefits into clinical effectiveness. Rather than drawing premature conclusions, these trials invite a more nuanced exploration of inflammatory profiles, treatment windows, and appropriate endpoints. Future research should adopt a precision medicine approach to fully assess the role of HA in cardiac surgery.

Abbreviations

HA

Heamoadsorption

CPB

Cardiopulmonary Bypass

ICU

Intensive Care Unit

IL

Interleukin

CABG

Coronary Artery Bypass Grafting

Re

CABG-Repeat Coronary Artery Bypass Grafting

AKE

Aortic Valve Replacement

MKE

Mitral Valve Replacement

MKR

Mitral Valve Reconstruction

RE

OP MKE-Repeat Mitral Valve Replacement

MIC

MKE-Minimally Invasive Mitral Valve Replacement

PCC

Prothrombin Complex Concentrate

PRBCs

Packed Red Blood Cells

FFP

Fresh Frozen Plasma

PCT

Procalcitonin

CRP

C-Reactive Protein

eGFR

Estimated Glomerular Filtration Rate

BUN

Blood Urea Nitrogen

AST

Aspartate Aminotransferase

ALT

Alanine Aminotransferase

Gamma

GT-Gamma-Glutamyl Transferase

SOFA

Sequential Organ Failure Assessment

ΔSOFA

Delta Sequential Organ Failure Assessment

ICU

Intensive Care Unit

Author contributions

All authors meet all four criteria for authorship recommended by ICMJE. All authors have seen and agree with the final contents of the manuscript. Ethical proposal: AH; Study design and conception: AH, AB Interviews: AB; Cardiotechnician training and conduction: DB; probe acquiring: DB, AB; Data management: NMO, AB; Statistics: NMO; data analysis: NMO, AB, AH; Manuscript drafting: AB, NMO; Manuscript revision: AB, NMO, AH, TA, PZ; Data interpretation: TA, AH, NMO, AB, Final approval of manuscript: AB, AH, TA, DB, NMO, PZ; Submission process: NMO, AB; Responsibility for concept: AH, NMO, AB.

Funding

CytoSorbents^® Europe GmbH supported the RECCAS trial by a grant for laboratory assays and compensated for the article processing charge of the initial protocol. No other funding was received for the study.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The Ethical Committee of Ruhr University Bochum, Germany, approved the prospective single-centre randomised controlled interventional trial RECCAS (ethical approval No. 5094–14), and patients were enrolled after written informed consent was provided by patients.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.