In the context of hepatic dysfunction, the removal of cytokines and hydrophobic, albumin‐bound hepatic toxins, such as bilirubin, bile acids, and amino acids, may have a beneficial effect on the clinical course of patients in liver failure [1]. Many extracorporeal liver support systems have been developed over the years, with the focus of removing these accumulating (mainly albumin-bound) toxins from the blood circulation, which has always represented a challenging removal target. To identify the most suitable principle to efficiently support the detoxification liver function, systems mainly based on albumin dialysis and plasma adsorption have been studied, including MARS, Prometheus, CPFA, and PAP [14,15,16,17,18,19]. More recently, a simple hemoperfusion system, CytoSorb, has demonstrated its ability to modulate inflammatory mediators, as well as bilirubin and bile acid levels [20,21,22].

Considering the limited comparisons available in the literature [23,24,25] we performed a retrospective comparative analysis on data collected with different extracorporeal liver support systems in our intensive care unit (Table 1). The objective was to evaluate the detoxification ability regarding TB, DB, and BA.

The main comparison was performed among patients treated with CytoSorb or CPFA due to similar treatment numbers, and the ability of both techniques to significantly adsorb the studied hepatic toxins. However, CytoSorb showed a significantly higher capacity expressed in term of MB. As shown in Table 2, the CytoSorb TB MB is five times more elevated than CPFA and three times for BA. This difference is confirmed when observing the evolution of MB over the treatment time (Fig. 2), and is also noticeable when considering adsorption in the first three hours of treatment, therefore giving a comparable treatment time among the two techniques. Indeed, TB MB was significantly higher for CytoSorb than CPFA (525.85 ± 35.84 vs 185.59 ± 30.62, p < 0.001), and so was BA MB (356.08 ± 25.62 vs 162.01 ± 24.69, p < 0.001).

The purification effectiveness was maintained by the CytoSorb device throughout the treatment duration, even if with less intensity toward the end and, importantly, no bilirubin release was observed. This was confirmed earlier in an in vitro study published by Gemelli et al. [20]. On the other hand, CPFA purification effectiveness tended toward zero at the end of treatment. Evaluation of the RR (Table 3) did not show significant results which is explained when considering the inability of this parameter to explain the effectiveness of an adsorption device. Indeed, MB is the only representative value to verify the purification effectiveness of one system as it is not affected by the continuous production of the molecules and ongoing release from the tissues, as is the case for RR.

Considering also the other techniques (Fig. 1)—even if the comparison is limited by treatment numbers—these results are confirmed, and CytoSorb showed the greater performance in term of MB (p < 0.05).

First of all, this higher capability might be explained by the elevated and available CytoSorb adsorption surface at the beginning of the treatment. Second, the different treatment durations certainly affect the total adsorption ability of the systems: CytoSorb is a system able to work up to 24 h, and in our experience, the median duration was 22 h (7, 26), whereas CPFA treatment was shorter because of technical and saturation issue, so that the median duration was 7 h (4, 14.5). This was the same for the other techniques which were shorter than the CytoSorb treatment (Table 1): MARS 6 h (4.5, 9.5), PAP 4.5 h (4, 5), and PROM 5 h (3.75, 7.5).

Other factors that could affect CytoSorb’s superior removal ability were the elevated TB and BA baseline concentrations, but this was valid for all adsorption therapies in general. The adsorption capacity of the CytoSorb cartridge is clearly dependent on the concentration of the target molecules, as it works in a concentration-dependent manner, efficiently removing high concentrations of target molecules with the goal of modulating the excess levels of toxic molecules, to regain control in complex situations. Remarkably, notwithstanding the higher baseline values, CytoSorb was able to significantly reduce TB, DB, and BA right up to the end of the treatment, reaching similar levels to that of CPFA (Table 3).

It is important to underline that both CytoSorb and CPFA seemed to be able to adsorb unconjugated bilirubin—a strongly albumin-bound molecule—together with direct bilirubin. This has already been demonstrated for CytoSorb [20] and is reiterated in our study considering the stability of the DB/TB ratio. This remained constant between baseline and end of the treatment values, confirmed by the DB/TB index around 1 (Table 2). Indeed, if more direct than unconjugated bilirubin were adsorbed, this index would have been different at baseline and at end of the treatment.

The ability to adsorb albumin-bound toxins might also explain the BA removal. BA are albumin-bound hepatic toxins, even if less tightly bound to albumin than bilirubin [25]. Unconjugated bilirubin presents an affinity binding of 9.5 × 107 M−1, unlike the two primary bile acids, cholic acid (CA) and chenodeoxycholic acid (CDCA), 5.5 × 104 M−1 and 0.3 × 104 M−1, respectively. The different composition of BA—and its affinity binding—may affect the removal efficiency of the systems, which are made from hydrophobic resin, and this could explain the minor adsorption obtained compared to the one for TB, even if we were not able to discriminate among the two BA types and understand their behavior. Nevertheless, the total BA adsorption was significantly superior with CytoSorb compared to CPFA (Table 2).

Technically, the experience with MARS, Prometheus, PAP, and CPFA underlines the need for careful management of anticoagulation, mainly heparin, to avoid clotting problems which affect the continuity of the treatment. One advantage of the CytoSorb system is its integration into the normal clinical practice, allowing also RCA. Indeed, the use of RCA guarantees excellent anticoagulation of both the entire extracorporeal circuit and the adsorbent system, maintaining its purifying effectiveness. Considering these benefits, the use of RCA during hepatic insufficiency could be used providing adequate precautions are taken to avoid the risk of citrate accumulation [26]. For example, there should be precautions taken regarding the limitation of blood flow, the use of a cut-off of ionized calcium in the extracorporeal circuit (at least up to 0.4 mMol/l), close monitoring of the total/ionized calcium ratio (which should not exceed the value of 2), and monitoring the values of pH, bicarbonates and lactates, whose increase must lead to the suspension of the RCA infusion by switching to another anticoagulation method.

Furthermore, CytoSorb can be easily inserted into an extracorporeal circuit for CRRT without changing the usual clinical routine, an important point, as renal failure is a frequent complication of liver failure. Therefore, the simplicity of CytoSorb use positively influences the continuity and duration of the treatment, not least the fact there are fewer complications in the set-up phase.

This study has some limitations. First of all, its retrospective nature and the relatively small and variable number of treatments and measurements per patient. However, considering the limited comparisons that include all the main extracorporeal liver systems noted in the literature, we considered it important to report our experience. Moreover, this study was focused on the analysis of the techniques effectiveness and not designed for clinical outcome evaluation. Further investigations are ongoing for this purpose.