The methods used to separate the plasma from the blood can be divided into centrifugation and filtration. Centrifugation is the older method, based on the separation of cellular elements from the plasma by rapid spinning, in which centrifugal force separates the different components according to their density, size, and molecular weight. This method has the advantage that there is no upper limit to the molecular weight of the substances to be separated out. It makes it possible to perform cytapheresis, in which cells of interest can be removed for therapeutic purposes or for later donations. The main drawback of centrifugation is the risk of thrombocytopenia. Moreover, it requires anticoagulation with citrate, so it can lead to hypocalcemia. Centrifugation is the method used by blood banks; it requires sophisticated difficult-to-transport equipment that limits its use in therapeutic apheresis in critical care environments.

In filtration, the cellular components of blood are separated from the plasma by passing the blood through a filter with large pores (0.2–0.7μm) that extracts molecules weighing up to 3millionDa. The mechanism of separation consists of applying pressure to transfer the blood across a synthetic membrane that is highly permeable due to the large size of its pores. This membrane is the central element of an extracorporeal circuit, similar overall to those used in intensive care units (ICU) for other purification treatments such as continuous renal replacement therapy (CRRT) or extracorporeal albumin dialysis with the molecular adsorbent recirculating system (MARS®). This approach requires a central venous catheter and anticoagulation with heparin. The advantages of filtration include the low risk of thrombocytopenia and the possibility of eliminating more plasma in less time. This approach also enables double filtration or cascade filtration in which the first filter separates the plasma, which is in turn passed through a second filter that has the capacity to selectively separate out certain molecules through filtration or adsorption.

Vascular access and blood flow through the extracorporeal circuit are fundamental for the success of the procedure. Vascular access can vary depending on the plasmapheresis technique, the condition being treated, and/or the duration of the treatment. In plasmapheresis by intermittent centrifugation and short-term procedures, peripheral venous accesses that provide blood flow of 50–90ml per minute can be used. In acute processes, the most frequently used accesses are temporary central venous catheters that provide blood flow of at least 70ml per minute, making it possible to complete the procedure in 3–4h. In conditions that require chronic treatment with plasmapheresis, permanent external vascular access or arteriovenous fistulas can be used. For plasmapheresis procedures indicated in critical patients by plasma filtration, where renal dysfunction and other organ failures are probably also present, temporary central venous catheters with a double lumen are the first choice for venous access, and they can also be used for other extracorporeal support techniques.

Apheresis procedures require anticoagulation to prevent clots from forming in the extracorporeal circuit. Citrate, unfractionated heparin, and hirudin are all used for anticoagulation, although unfractionated heparin has long been the anticoagulant of choice for filtration plasmapheresis in ICUs. It is usually administered with an initial intravenous bolus of 40–60UI/kg followed by continuous infusion of 20UI/kg/h to maintain the activated partial thromboplastin time (aPTT) between 180 and 220s throughout the treatment.6 These doses are usually higher than those required for CRRT because a significant portion of the heparin is extracted together with the plasma. It is important to note that the loss of coagulation factors derived from the negative balance between the plasma extracted and the replacement fluid can result in a greater anticoagulant effect than might otherwise be predicted. In patients with a high risk of bleeding, the dose of anticoagulants must be considerably lower. Other authors consider that the administration of anticoagulants during plasmapheresis by membrane filtration is unnecessary.7 No published studies have looked at the risk to benefit ratio of the common practice of circuit anticoagulation, therefore further research should be conducted. In recent years, regional anticoagulation with citrate has been proposed as a new therapeutic strategy to maintain the permeability of the extracorporeal circuits, preventing early clotting and minimizing the systemic effects in the patient.8

The characteristics of the replacement fluid will depend on the type of disease for which the treatment is being done. The volume of plasma extracted is replaced with a replacement solution with an appropriate electrolyte composition and colloid osmotic activity. The volume of replacement fluid must always be the same as that of the effluent obtained. In adults, the effluent can be replaced with crystalloid solutions only if the volume extracted is less than 1000ml. When a larger volume is extracted, it is essential to use colloids, with 4–5% human albumin being the colloid of choice. For this purpose, 20% human albumin is diluted with crystalloids or polyelectrolyte solutions. Currently, pasteurized liquid plasma proteins have the advantage of being sold in 500ml bottles, which are more economical and do not need to be manipulated; for these reasons, they have practically supplanted diluted 20% albumin as a replacement fluid.9 In patients with thrombotic thrombocytopenic purpura, coagulation factor deficiencies, or immune deficiencies, fresh frozen plasma must be used. Supplementation with intravenous immunoglobulin after plasmapheresis has been advocated to counteract progressive immunoglobulin depletion, but this provides only transient increases in levels, and is of questionable benefit.

Table 3 lists the advantages and disadvantages of the different replacement fluids.

The citrate administered as an anticoagulant for plasmapheresis bonds with calcium and can lead to symptoms of hypocalcemia. Likewise, fresh frozen plasma contains a high proportion of citrate,14 so if it is administered as a replacement fluid, it can cause the same effect. Administering albumin as a replacement fluid can also lead to hypocalcemia due to direct calcium sequestration.14 The symptoms include paresthesias around the mouth and extremities, dizziness, muscle cramps, nausea, and vomiting. Severe cases can lead to prolongation of the QT interval, arrhythmias, chest pain, seizures, and hypotension. To reduce the incidence of these complications, intravenous calcium can be administered prophylactically and/or the amount of citrate perfused can be reduced.15 Although there are many guidelines for the administration of intravenous calcium,11,16,17 one simple approach is to administer a 10ml bolus of 10% calcium gluconate every hour during the plasmapheresis procedures.14

Administering fresh frozen plasma as a replacement fluid can give rise to anaphylactic reactions such as fever, stiffness, urticaria, pruritus, bronchospasm, hypotension, and laryngeal edema.14 Anaphylactic reactions to albumin are much rarer; they can be associated with the formation of antibodies to polymerized albumin or they can develop in patients on angiotensin-converting enzyme (ACE) inhibitors.15 Most anaphylactic reactions are mild to moderate; only 0.1% of cases are classified as severe.13 Due to the relatively high incidence of anaphylactic reactions, patients who need plasmapheresis with massive fluid replacement with fresh frozen plasma (e.g., those with thrombotic thrombocytopenic purpura) are often pretreated with 50mg of intravenous diphenhydramine.21 In patients with previous reactions to fresh frozen plasma in whom fresh frozen plasma must be used as the replacement fluid (e.g., those with thrombotic thrombocytopenic purpura), prophylaxis can be administered, with 50mg oral prednisone 13h, 7h, and 1h administered before the plasmapheresis session, 50mg oral diphenhydramine administered 1h before the session, and 25mg ephedrine administered 1h before the session.22 If a severe reaction occurs, with refractory hypotension, severe bronchospasm, or laryngeal edema, the usual medical treatment and ICU support for anaphylactic shock must be administered. ACE inhibitors must be suspended at least 24h before any plasmapheresis procedure.23

When albumin is the replacement fluid, the risk of an infection associated with plasmapheresis is due to the depletion of immunoglobulins (Ig). Replacing plasma with albumin results in a 60% decrease in Ig levels, and multiple plasmapheresis sessions in short time periods will lead to a drop in Ig levels that can persist for several weeks.21 Given that the depletion of Ig can worsen the patient's ability to fight infection, it is recommendable to restore the normal Ig levels with the intravenous infusion of 400mg/kg in patients who develop severe infections in the period after plasmapheresis.15 When fresh frozen plasma is the replacement fluid, the risk of an infection associated with plasmapheresis is due to viral transmission. The estimated risk of transmission is 1– 2 for every million units transfused for the human immunodeficiency virus and for the hepatitis C virus, and 1 for every 200,000–500,000 units transfused for the hepatitis B virus.24

Transfusion-related acute lung injury (TRALI) is characterized by the development of acute respiratory failure with non-cardiogenic pulmonary edema, often accompanied by hypotension that appears abruptly during the transfusion of blood products or within hours after the procedure. It is caused by the presence of antibodies (Ab) in the fresh frozen plasma infused and their reaction with antigens (Ag) in the patient's white blood cells. The Ag–Ab complex results in neutrophil activation and cytokine release, leading to increased endothelial permeability. There are only a few reports on development of this severe complication in small case series, therefore, the frequency and risk factor for TRALI in critically ill patients during plasmapheresis still remain to be determined in large prospective studies.25

The incidence of hypotension varies (Table 4). Various factors can cause hypotension in plasmapheresis, including inadequate volume of replacement fluid (Figure 1), vasovagal episodes, anaphylactic reactions to different substances (replacement fluids, presence of anti-IgA antibodies in patients with IgA deficit, biocompatibility of membranes used in the plasma filter, sensitivity to ethylene oxide), arrhythmias induced by hypocalcemia and/or hypokalemia, generation of bradykinins in patients on ACE inhibitors, bleeding (related to the underlying disease, to the use of heparin as an anticoagulant, or to depletion coagulopathy), TRALI, pulmonary embolism, cardiovascular collapse, or factors related to the underlying disease, such as autonomic dysfunction in patients with Guillain–Barre syndrome.15

Figure 1.

Changes in blood volume in plasmapheresis with albumin replacement fluid using an online volume monitor. Variation in blood volume (Crit-Line®). Starting plasmapheresis in a patient with a myasthenic crisis and replacement with 5% albumin solution. There is an abrupt drop in blood volume (about 10%), probably mediated by the transfer of fluids from the vascular space to the interstitial space due to the drop in the colloid osmotic pressure.

(0.11MB).

Like polymyxin E (colistin), polymyxin B belongs to the group of cationic polypeptide antibiotics. Both have efficacious antimicrobial activity against gram-negative bacilli; however, nephrotoxicity and neurotoxicity are important limitations for both. Polymyxin B is characterized by strong bonding with circulating endotoxin (1:1). Polymyxin B has been incorporated into polystyrene and polypropylene fibers in a device (Toraymyxin®) that has a high capacity for the adsorption of endotoxin. Since the drug does not enter the systemic circulation, its secondary effects are not problematic. Hemoperfusion treatment with polymyxin B takes 2h (saturable mechanism), and it is recommended to carry out two treatments in two consecutive days. Treatment should be started early in cases of elevated endotoxemia. The first studies of this device done in Japan, where it was invented, found decreased levels of endotoxin in plasma, improved hemodynamics, and a decrease in mortality. In 2005, the first European multicenter randomized prospective study was published.29 This study included 35 patients (17 in the polymyxin B group and 18 controls) with septic shock secondary to intraabdominal infection. A single session of hemoperfusion improved the cardiac index but did not significantly reduce the levels of endotoxin in plasma, SOFA scores, ICU stays, days on mechanical ventilation, or 28-day mortality. In 2007, a meta-analysis30 of 28 publications (9 randomized controlled trials, 7 non-randomized parallel studies, and 12 pre-post design studies) including a total of 1425 patients, of whom 978 had received treatment with polymyxin B, found that Toraymyxin® treatment was associated with a 21.2pg/ml decrease in the concentration of endotoxin (95% CI: 17.5–24.9pg/ml; p<0.001), a 19mmHg increase in mean arterial pressure (95% CI: 15–22mmHg; p<0.001), a 1.8μg/kg/min decrease in the dose of vasoactive amines (95% CI: 0.4–3.3μg/kg/min; p=0.01), a 32-unit increase in the Pa02/Fi02 ratio (95% CI: 23–41 units; p<0.001), and a halving of the 28-day mortality risk (RR 0.53; 95% CI: 0.43–0.65; p<0.001). However, in general, the methodological quality of the studies included was poor (Jadad scale<3) due to the low number of patients included, incorrect randomization, and/or lack of double blinding. Two years later, the EUPHAS study was published.31 In this multicenter randomized controlled trial, patients with septic shock secondary to abdominal infection who underwent emergency surgery were randomized within 6h of the intervention to receive either conventional treatment (30 patients) or conventional treatment plus polymyxin B hemoperfusion (34 patients) administered in two 2-h sessions in two consecutive days. The results showed improved mean arterial pressure with a reduction in the need for vasopressors, improved SOFA score, and a decrease in 28-day mortality (53% in the conventional treatment group vs. 32% in the polymyxin B group; adjusted HR 0.36: 95% CI 0.16–0.8). The ethics committee stopped the study early after the intermediate analysis showed decreased mortality in the treatment group. The major limitations were that the trial was not blinded, endotoxin activity was not measured, and likelihood ratios was used instead of Fisher's test, the preferable approach to estimating mortality. In 2013, Zhou et al.32 published a meta-analysis of the efficacy of blood purification treatments in patients with sepsis that showed a positive effect on survival in patients receiving hemoperfusion with polymyxin B. However, these results were not corroborated in another study based on propensity score matching, which found no improvement in survival in surgical patients with septic shock secondary to abdominal infection.33 Another randomized controlled trial, ABDOMIX,34 published in 2015, studied patients with septic shock due to peritonitis who required surgical intervention. A large number of patients were included in each arm (119 treated with hemoperfusion with polymyxin B vs. 113 controls), but no differences in 28-day mortality were found between groups. However, many patients received only one treatment and there were many cases of filter clotting, so we can deduce that polymyxin B was inadequately administered in a large proportion of patients in the treatment group. Another randomized controlled trial, EUPHRATES,35 is currently recruiting patients with septic shock and high endotoxin levels (EAA>0.6) in 50 ICUs in the USA and Canada. Its results will probably provide valuable information about the efficacy of hemoperfusion with polymyxin B. Until the results are available, following the indications proposed by Candel et al.,36 hemoperfusion with polymyxin B should be considered in patients with septic shock due to abdominal or urinary infections with gram-negative bacteria that do not improve within 6–12h after appropriate resuscitation and standard active and aggressive treatment.

This device has two porous polyethylene discs covered with a synthetic peptide that has a high capacity for the adsorption of endotoxins. To date, most publications about this device have been observational studies reporting some positive results but including few patients.37,38 Yaroustovsky et al.39 compared the LPS Adsorber versus hemoperfusion with polymyxin B in a small number of patients with sepsis due to gram-negative bacteria and found no differences in the outcomes achieved with the two approaches. In another study, Adamik et al.40 observed that the LPS adsorber effectively eliminated endotoxins in patients with septic shock, resulting in significant improvements in hemodynamic parameters and organic dysfunction, although without repercussions in the ICU stay or mortality. Further studies are necessary to determine the efficacy of this device.

This polysulfone and polyacrylonitrile (AN-69) filter has ample adsorption of inflammatory cytokines and endotoxin, but clinical experience is limited given the scant number of studies published.41

Based on endotoxin's affinity for human albumin, this system incorporates albumin in a polymethacrylate filter. Although the in vitro studies were promising, the clinical studies have focused mainly on safety and tolerance and have yet to be able to demonstrate clinical efficacy.42

Since Ronco et al.43 found improved survival in critical patients with acute renal failure treated with high rates of ultrafiltration (≥35ml/kg/h), there has been a trend toward using high filtration volumes in patients with severe sepsis/septic shock with the aim of affecting the levels of circulating cytokines, the systemic inflammatory state, and the hemodynamic situation.

Later studies found benefits with higher ultrafiltration rates, so after the Pardubice consensus conference,44 high volume hemofiltration (HVHF) was redefined as continuous high volume treatment with an effluent rate between 50 and 70ml/kg/h for 24h per day or intermittent treatment with an effluent rate between 100 and 120ml/kg/h for 4–8h followed by hemofiltration at conventional doses.

On the other hand, it has been postulated that applying HVHF in the early phases of sepsis could eliminate or reduce the peak levels of proinflammatory and anti-inflammatory cytokines. This elimination of cytokines from the bloodstream would create a gradient that would favor the extraction of these cytokines from the tissues, thus limiting damage to organs.45,45,46 In a meta-analysis, Lehner et al.47 point to the possible usefulness of HFHV pulses>50ml/kg/h in reducing the levels of circulating cytokines, and some of the studies included also found a greater decrease in the doses of vasoactive amines.

Nevertheless, other authors48,49 found no benefits in survival or in renal function with doses greater than 25–30ml/kg/h. A recent multicenter clinical trial that randomized patients to receive either HVHF (70ml/kg/h) or standard hemofiltration (35ml/kg/h) found no significant differences in mortality, improvements in hemodynamics or organ dysfunction, length of mechanical ventilation, time requiring CRRT or recovery of renal function, ICU or hospital stay, or adverse effects attributable to the technique.50 These findings are in line with those obtained in another later systematic review and meta-analysis.51

In summary, the heterogeneity of the studies hinders their inclusion in reviews and meta-analyses, so few of these types of studies have been done. Not enough evidence has been published to recommend the systematic use of HVHF at the reported doses in septic patients with acute renal failure.

Treatment with coupled plasma filtration and adsorption (CPFA) comprises 3 phases: (1) Plasma filtration. The patient's blood is passed through a filter that allows the components of plasma to cross the membrane while the cells are returned to the patient. (2) Purification. Next, the plasma passes through an absorbent cartridge composed of a resin that enables the absorption of inflammatory mediators and endotoxins, and then the treated plasma is returned to the circulation. (3) Hemofiltration or hemodialysis to eliminate water and low-molecular-weight toxins. This is made possible by a second filter.

Studies done in vitro have shown the efficacy of CPFA in adsorbing interleukins and TNF-α.52 They have also observed a decrease in the need for vasoactive amines in patients with septic shock,53 and CPFA has proven superior to HVHF in eliminating inflammatory mediators in septic patients with multiple organ failure.54 A recent prospective multicenter study was stopped for futility after the investigators were unable to find a decrease in mortality in patients with septic shock treated with CPFA.55 Nevertheless, a subgroup analysis showed that hospital mortality decreased in patients with greater volumes of plasma treated, a finding that was also suggested in another study in patients with septic shock and multiple organ failure.56 Evidence for CPFA in sepsis is sparse; studies are heterogeneous, include few patients, and suffer from various methodological shortcomings, so further research is required.

Membrane cutoff is defined as the mean value of the molecular weight of molecules with a sieving coefficient of 0.1. Clinically, this point is defined by the molecular weight of the largest molecule that passes through the membrane (≥60.000Da). Morgera et al.57 found that hemofiltration with high cutoff membranes was superior to conventional hemofiltration in the elimination of IL-6 and IL-1ra and in the reduction of doses of noradrenalin in septic patients with acute renal failure. Later reviews found a significant improvement in hemodynamic parameters and in oxygenation indices in septic and non-septic patients.58 Using convection techniques with this type of membranes has proven to be an efficient way to reduce the levels of circulating cytokines, although these techniques are associated with high albumin washout. The exclusive use of diffusion seems to be the most appropriate option for increasing cytokine elimination and reducing excessive albumin loss.59

Finally, Table 5 describes other extracorporeal depurative techniques with limited clinical experience.