Cytokine clearance with CytoSorb® during cardiac surgery: a pilot randomized controlled trial

This prospective single-centre randomized controlled trial took place in the Centre Hospitalier Universitaire Vaudois (CHUV), Lausanne, Switzerland, between May 2016 and January 2018. The target population included patients planned for elective cardiac surgery with expected long CPB duration and deemed at high risk of post-operative complications. The study protocol was approved by the Ethics Committee Vaud (2015-00010) and registered at ClinicalTrials.​gov (NCT02775123).

Baseline pre-operative characteristics (age, sex, body weight and chronic diseases), chronic medications (aspirin, angiotensin-converting enzyme inhibitors and angiotensin II receptors blockers), routine laboratory values and data relevant to the undertaken surgical procedure (type, CPB characteristics and fluid balance) were collected. To be considered, pre-operative values had to be obtained within 24 h of the procedure and discharge values had to be the latest obtained before discharge. Post-operative values for haemoglobin, platelets, aPTT and PT were obtained on ICU admission. Outcomes’ data were collected at ICU and hospital discharge.

General anaesthesia was induced with propofol, sufentanil and rocuronium. It was maintained with sufentanil boluses, sevoflurane before and after bypass and propofol during bypass. Tranexamic acid was administered as a 10-mg/kg bolus on anaesthesia induction followed by a 1-mg/kg/h infusion throughout CPB. Heparinization was initiated with a 300-UI/kg bolus and titrated according to haemostasis management system (HMS) Plus (Medtronic, Minneapolis, USA). An ACT > 400 s was required to initiate CPB. At the end of the procedure, heparin was reversed with protamine titrated according to HMS calculations.

The CPB included a Capiox® FX25 (Terumo, Tokyo, Japan) or Quadrox-i® (Maquet, Rastatt, Germany) membrane oxygenator and a heater-cooler system (Livanova, London, England). A roller or centrifugal pump was used according to the perfusionist’s preferences. The circuit was primed with heparin and Plasmalyte-A® (Baxter, Deerfield, USA). If deemed necessary, a haemofilter (haemoconcentrator, Sorin) was used. A non-pulsatile flow at 2.4 L/m²/min was maintained with a target mean arterial pressure between 60 and 80 mmHg. For this purpose, phenylephrine was administrated if deemed necessary. Myocardial protection was achieved using either St. Thomas® (Bichsel Laboratorium, Interlaken, Switzerland) or Custodiol® HTK (Sandor Medicaids, Hyderabad, India) cardioplegia solution, according to the surgeon’s preferences. Transesophageal echocardiography (TEE) was performed throughout the procedure.

For patients allocated to the HA group, a CytoSorb® cartridge was purged with NaCl 0.9% and integrated into the CPB circuit. The device was inserted in a side arm connected to the outflow line (high-pressure section) and the venous reservoir, prior to the oxygenator. No flow monitoring was conducted.

Blood samples were drawn in ethylene-diamine-tetra-acetic acid (EDTA) (cytokine analyses) or 0.106 M sodium citrate (coagulation factors) tubes. All samples were collected using standard hygiene precautions from patients’ arterial line or within the CPB circuit (T1 pre and post-adsorber measurements).

EDTA samples were centrifuged at 1500g at 4 °C for 10 min. Three aliquots of 100-μl plasma were prepared and stored at − 80 °C. Cytokine quantification was performed by the Mouse Metabolic Evaluation Facility (MEF), University of Lausanne, using Human Cytokine Magnetic 10-Plex Panel on the Luminex® Platform (Thermo Fischer Scientific®, Waltham, USA) [17]. Detection thresholds for cytokines are presented in Additional file 1: Table S3.

Based on previous literature [18], we calculated that, in order to be able to demonstrate a 25% change in serum IL-6 concentration, assuming a mean value of 200 pg/ml and a standard deviation of 50 pg/ml, we would need to recruit 15 patients in each arm to achieve a power of 80% with statistical significance set at 0.05.

A continuous variable is reported using mean and standard deviation (SD) or median and interquartile range (IQR) according to data distribution. Categorical variables are described using absolute frequencies and relative percentages.

For baseline characteristics and clinical outcomes comparisons, differences between groups were analysed by using t test or Wilcoxon signed-rank test for continuous variables and Fisher’s exact test for categorical variables. A p value < 0.05 was considered statistically significant.

For cytokine level comparisons, since cytokine levels were non-normally distributed, they are reported as median (IQR). The Mann-Whitney U test was performed for inter-group comparisons at each time point. Effect of time within each group was assessed using the Kruskal-Wallis one-way analysis of variance. A p value < 0.05 was considered statistically significant.

For coagulation factors’ analyses, comparisons of pre- and post-adsorber measurements of coagulation factors’ activity were performed using paired Student’s t tests. The effect of HA on coagulation factors throughout study time points was assessed by repeated measures’ analysis of variance (ANOVA) and analysis of covariance (ANCOVA) models. Final ANCOVA models included FFP administration, fluid balance and baseline value. Statistical threshold was determined as p = 0.016 after Bonferroni correction.

All analyses were performed using IBM SPSS Statistics (25).

Baseline and peri-operative characteristics of included patients are detailed in Table 1. There was no difference between the two groups in terms of baseline or peri-operative characteristics except for a lower protamine/heparin ratio in the HA group (p = 0.02). Median CPB duration was 142.5 (114.3–200) min.

Cytokines’ measurements throughout study time points are presented in Fig. 1. Il-6 and Il-4 baseline serum levels were higher in the HA group. Compared to baseline values, the surgical procedure was associated with an increase in IL-6 (peak level 6 h after CPB initiation, T3), MCP-1 and IL-10 levels (peak level at the end of CPB, T2). On the other hand, it was associated with a decrease in IL-1α, IL-1β, IL-2, IL-4, IL-5 and IFN-γ levels (nadir at T2 and T3). No change in TNF-α levels was observed until T4. There was no statistically significant difference between the two groups in serum levels of any of the cytokines of interest at any time point.

Clinical outcomes are reported in Table 2. There was no significant difference between the two groups in the use of vasoconstricting drugs (p = 1.0), mean noradrenaline dose on the first post-operative day (p = 0.87), use of any post-operative inotrope (p = 0.68), duration of mechanical ventilation (p = 0.31), incidence of AKI (p = 1.0), ICU length of stay (p = 1.0) and ICU and hospital mortality (both p = 1.0).

We performed a single-centre pilot randomized controlled trial on 30 patients undergoing elective cardiac surgery and deemed at high risk of peri-operative complications. In those patients, we evaluated the safety and efficacy of CytoSorb® haemoadsorption during CPB in comparison with standard management. We found that the procedure, as described, appeared safe and feasible as no immediate or delayed complication attributable to the device was observed. A thorough investigation of coagulation parameters did not demonstrate any relevant alteration of coagulation factors or thrombocytes levels associated with the procedure. However, the intervention was not associated with a decrease in key cytokine levels. Similarly, although the study was not powered to examine these outcomes, it was not associated with a difference in terms of need for vasoconstrictors, post-operative AKI or need for RRT, ICU length of stay and in-hospital mortality.

Our main result is consistent with findings from a previous and similar RCT conducted in Austria by Bernardi et al. [15]. In this trial, patients undergoing elective cardiac surgery with an expected CPB duration of more than 120 min were randomly allocated to either CytoSorb® HA or control group. After the exclusion of five patients, 32 were included in the main analysis (16 in each group). The authors did not find any difference in peri-operative levels of IL-6, IL-10, IL-18, IL-1β and TNF-α and HMGβ1 except for a longer decay for IL-10 in the CytoSorb® group.

Patients included in both trials appear to be comparable in terms of age, EuroSCORE and comorbidities. Although arguably patients included in our study underwent more complex procedures (aortic procedures), this did not translate in a longer CPB duration (138/145 min versus 170/191 min). Hence, both studies have evaluated CytoSorb® HA in a collective of patients at high risk of complications.

We have analysed a different panel of cytokines, including pro-inflammatory IL-1α, IL-1β, IL-6, IFNγ, TNFα and MCP-1, as well as anti-inflammatory, IL-4, IL-5 and IL-10. However, similar to Bernardi et al., we failed to demonstrate any effect of CytoSorb® HA on these cytokines. Of note, we did not observe higher IL-10 levels in the CytoSorb group. The absence of efficacy observed in the Austrian trial [15] is therefore confirmed by our data. In addition, consistent with the observation made, we have observed large inter-patient variability.

The observed lack of efficacy of the therapy might be explained by the low inflammatory response observed in patients included in both trials. Indeed, in an observational study, intra-operative CytoSorb HA was associated with a decreased requirement for post-operative vasoconstrictors in 39 patients with infective endocarditis compared with historical controls [13]. However, in these patients, peak (post-operative) IL-6 levels were around 400 mg/ml, a value much higher than those observed in our trial (median peak IL-6 levels (2 h post-CPB) 120.8 [49.0–160.8] pg/ml in the HA group and 118.7 [68.4–255.9] pg/ml in the control group).

To the best of our knowledge, this is the first trial evaluating the consequences of CytoSorb® HA on coagulation parameters. Several studies have reported a trend for thrombocytopenia [19] during HA. We have not observed such association.

This study has several strengths. Despite the small sample size, baseline characteristics were well balanced between the two groups of patients. We have examined a large panel of pro- and anti-inflammatory cytokines at different relevant time points. In addition, we have conducted a thorough evaluation of CytoSorb® effect on several coagulation parameters. These analyses involved cross-adsorber clearance measurement, coagulation factors’ levels throughout relevant time points and conventional haematological evaluation. To the best of our knowledge, our study is the first to provide such a broad assessment during CytoSorb® HA.

Some limitations are however present. First, it is a monocentric pilot study, on a rather small collective of patients. Our results are therefore subject to selection bias, and the external validity of our results can be limited. However, our findings are largely consistent with those obtained in a similar study but conducted in another health care system and another country.

Second, the duration of CytoSorb® HA was restricted to the CPB duration, which might not be sufficient to demonstrate efficacy. Indeed, cytokine release requires complex intracellular signalling mechanisms before they can be upregulated, expressed and secreted. The device insertion might have been too early relative to CPB-induced immune response and the duration of the therapy insufficient. However, insertion of the cartridge during CPB is a pragmatic, easy and potentially generalizable practice with minimal manipulations and risks to the patient. Longer therapy would require insertion of a double-lumen catheter and prolonged anticoagulation, which might not be desirable.

Third, although we have selected patients at risk of post-operative complications, the actual rate of such complications was relatively low and most patients required little ICU support. This might be consistent with the relatively low cytokine levels. A positive effect of the device on patients with higher inflammatory response and higher incidence of post-pump syndrome cannot be ruled out.

Finally, the utilization of modified ultrafiltration in some patients might have attenuated the effect of the device [20]. However, such utilization was limited to two patients in each group. Post hoc analyses performed after exclusion of these patients confirmed main results.

Our study shows that, even in patients at high risk of complications, as defined by clinical criteria, CytoSorb® HA, during CPB is not associated with a significant decrease in cytokine levels. This absence of effect might be related to a low inflammatory response, to an insufficient therapy duration or to an inadequate timing relative to pro-inflammatory cytokine production. Although an effect in patients with higher rate of post-pump syndrome and higher inflammatory response cannot be ruled out, CytoSorb® HA is not likely to be beneficial in the vast majority of elective cardiac procedures with low to moderate inflammatory responses. Together with the trial from Bernardi et al. [15], and despite small sample sizes, we can now conclude that routine application of CytoSorb® HA seems not to be justified for elective cardiac procedures. Indeed, given the absence of effect in two well-conducted RCT in patients with high risk of complications and prolonged CPB but nevertheless low cytokine levels, it is highly unlikely that a larger or multi-centre trial would demonstrate a benefit.

Further studies are required to evaluate its potential in particular situations perhaps associated with higher cytokine levels such as emergency procedures, acute infectious endocarditis or heart transplantation as suggested by observational studies [12, 13]. Post-operative use of CytoSorb might also need to be evaluated [20]. The observed heterogeneity of peri-operative measurements might suggest that some particular patients could benefit from the therapy. Hence, further studies might attempt to include a point-of-care evaluation of cytokine levels to enrich study population and restrict the intervention to patients with elevated cytokine levels.

On the other hand, our trial has provided important safety data in particular regarding the coagulation system. The absence of significant alteration of the coagulation system during the therapy is of major interest. The small but significant decrease in AT and FII across the adsorber might be interpreted as coagulation activation. Together with the decreased protamine to heparin ratio, reflecting known heparin adsorption in the HA device and the decrease FXII activity at the end of CPB reflecting contact activation, our findings further validate the need for therapeutic anticoagulation associated with the procedure.