The Prosorba column acts as a selective plasma exchanger [88]. Protein A, a major cell wall component of Staphylococcus aureus, binds human immunoglobulin (Ig) G with high affinity. In Prosorba columns, protein A is covalently bound to a silica matrix and used to purify the patient’s plasma. Cells and plasma are separated with a continuous cell separator and 1,250 ml plasma passes through the column. Treated plasma is then reconstituted with cells and returned to the patient. This procedure was initially approved for treatment of idiopathic thrombocytopenic purpura. Later, Prosorba immune-adsorption was successfully used in rheumatoid arthritis [88] and was granted FDA approval for treatment of moderate to severe forms of this disease. Few reports discuss the use of apheresis in sepsis [89].

The mechanism underlying Prosorba-induced improvement in rheumatoid arthritis and sepsis is unknown. Obviously, direct removal of IgG plays no role because as little as 1.5 % of circulating Ig is removed, and levels of rheumatoid factor and circulating immune complexes are barely modified throughout the entire treatment cycle [90]. Additional factors, such as complement activation or production of anti-idiotype antibodies, likely do contribute to the observed clinical improvement [90]. Recently, Suda and coworkers developed a new type of sorbent that specifically targets HMGB-1 [91].

As for other sorbents and other related techniques, antimicrobial adsorption can be very high [92, 93] as described also for hyper-adsorptive membranes [94].

CRRT in non-septic AKI can be is generally applied using a non-adsorptive biocompatible membrane of 1 m². However, some relatively cheap non-selective adsorptive membranes such as the AN69 ST type are already used on a large scale in this condition as well.

Regarding HVHF and after the two large randomized trials that are negative regarding mortality end points, HVHF is no longer recommended in clinical practice but may still be evaluated in appropriate trials.

The IVOIRE trial did suggest that an early start at AKI RIFLE injury score in septic shock with AKI may improve mortality as compared to a late start at AKI RIFLE failure score. Delay of therapy between ICU admission and CRRT start above 48 h may also impact negatively mortality according to IVOIRE. These data need to be further confirmed or not by randomized trials.

High cut-off membranes can remove cytokines and are under evaluation for treatment of rhabdomyolysis and sepsis. However, a cut-off of 60,000 Da enhances albumin loss. Instead, medium cut-off membranes (e.g. cut-off at 50,000 Da) should be preferred.

To increase removal of inflammatory mediators in septic AKI, the use of highly adsorptive filters that can be considered in CRRT such as the AN69 ST, septeX, PMMA, and Oxiris membranes but without strong clinical evidences.

The CytoSorb membrane binds cytokines and is seems a promising membrane for sepsis treatment, though it fails to capture endotoxin and IL-10. So far, there is more evidence for HCO and PMX filters. The AN 69 Oxiris membrane may be a valid alternative as it does capture both endotoxin and cytokines. AN69 ST also adsorbs the important upstream mediator HMGB-1. PMMA is a powerful scavenger of endotoxin and numerous other cytokines. Finally, high porosity septeX membranes may perform well especially when used in CVVH mode.

Clinical evidence for the benefit of adsorptive columns and selective plasma exchange is still scarce.

More studies are definitely needed to identify the ideal membrane or sorbent for adjuvant treatment of sepsis.

The highly adsorptive membranes AN69 ST and PMMA may capture significant amounts of crucial antibiotics such as colistin, amikacin, and vancomycin necessitating dose adaptation.

Heparin-coated membranes do not preclude the use of anticoagulation.