In extracorporeal albumin dialysis (e.g. MARS—molecular adsorbent recirculatory system; Gambro, Sweden) , the patient’s blood is drawn off and passed through a cartridge containing a semi-permeable, albumin impregnated polysulfone membrane. The dialysate (20 % albumin) is also passing through this cartridge, on the opposite side of the membrane. The transfer of albumin-bound molecules (molecular weight < 50 kDa) therefore occurs across this membrane from the high-concentration compartment (patient’s blood, saturated binding sites on albumin molecules) to the low concentration compartment (dialysate, empty binding sites on albumin molecules) via the intermediate process of binding to membrane-bound albumin . This is possible due to a greater affinity of membrane-bound albumin for toxins. Fresh dialysate continually replaces that in the cartridge to maintain the concentration gradient, while the toxin-laden dialysate moves on to a standard haemofiltration cartridge where the removal of water-soluble substances occurs in standard fashion (Fig. 26.1). Importantly, prior to returning to the first filter, the post-haemofiltration fluid must pass through adsorption columns containing activated charcoal and ion exchange resins. These remove albumin-bound, non-water-soluble toxins and allow the post-haemofiltration fluid to return to the first filter as fresh albumin dialysate.

A simpler, non-commercial version of this which dispenses with recirculation of the dialysate is single-pass albumin dialysis (SPAD). In this case, 2–5 % albumin is used as dialysate and the exchange of protein-bound toxins occurs across a high-flux semi-permeable membrane (albumin does not cross the membrane). The dialysate is discarded after a single pass. Haemofiltration may also be added to enhance clearance of water-soluble material.

Alternative systems use fractionated plasma separation and adsorption (FPSA). For this, a membrane with a larger cutoff of 250–300 kDa is used, allowing albumin and its bound toxins to pass across into a separate circuit . The fluid in this second circuit is passed through adsorption columns (neutral resin adsorber and anion exchanger), purging the patient’s albumin of toxins, and in a sense regenerating it before returning it to the blood from whence it came. In the Prometheus system (Fresenius Medical Care AG, Bad Homburg, Germany), this is combined with high-flux haemodialysis of the patient’s blood (Fig. 26.1).

In selective plasma filtration technology (SEPET, Arbios Systems Inc., Los Angeles, CA), a large-pore blood/plasma filter selectively filters and then discards the plasma fraction containing molecules of molecular weight < 100 kDa (therefore including albumin). The lost fluid is replaced with an electrolytic solution, 5 % albumin and fresh-frozen plasma. The retained components of fluid include clotting factors, immunoglobulins, complement proteins, and stimulators of hepatic regeneration [14].

In plasmapheresis, a well-established modality used for the treatment of many autoimmune disorders, the patient’s blood is treated so that plasma is separated out from cellular components. This plasma is discarded and replaced by donor fresh-frozen plasma and/or albumin. Clearly, this will not only result in the loss of the patient’s albumin and any toxins in the plasma but also clears other components in the plasma fraction, such as pro-inflammatory cytokines and circulating antibodies.

Data on bio-artificial systems are sparse. To date, only one multi-centre randomised controlled trial in ALF has been conducted using HepatAssist–BAL [15] ) . The treatment was well tolerated with few side effects but for thrombocytopenia, and found reduced levels of bilirubin (but not other metabolic factors) in the BAL group. However, the trial was stopped prematurely due to the low likelihood of a significant treatment effect on 30-day mortality (Table 26.2). Subgroup analysis suggested a possible beneficial effect in patients with fulminant/subfulminant hepatic failure. Extracorporeal liver assist device (ELAD) has also been trialed in ALF in much smaller numbers (phase I and II trials only) and has demonstrated safety but did not demonstrate any significant outcome benefits (Table 26.2) [16]. The ELAD device has been modified since this trial to include a greater mass of hepatocytes, a bigger membrane pore size, and a greater cartridge flow rate.