FormalPara Take-home message

Therapeutic plasma exchange (TPE) procedures performed by trained personnel are a safe and effective therapeutic approach for patients suffering from diseases listed in the guidelines of the American Society for Apheresis.

The creation of a specific registry for TPE administered in the intensive care unit would allow for a robust database to assess efficacy and safety of TPE in critically ill patients.

Introduction

Therapeutic apheresis encompasses the removal of plasma (plasmapheresis) or blood cells (cytapheresis, i.e., erythrocytes, leukocytes, or platelets) from the patient’s blood. If plasma is removed not for donation but for therapeutic purposes and is replaced by donor plasma, colloid, or crystalloids or a mixture thereof, it defines therapeutic plasma exchange (TPE) (Fig. 1). TPE serves to remove pathogenic substances (e.g., autoantibodies or toxic agents) and/or to administer deficient substances present in plasma of healthy donors (e.g., a disintegrin and metalloproteinase with a thrombospondin type 1 motif, member 13, ADAMTS13) though other potential immunomodulatory effects may be involved [1]. The indications for TPE have been refined over time. Many patients who require TPE are critically ill needing admission to the intensive care unit (ICU). TPE is an invasive procedure with often emergent indications, demanding its execution as soon as possible. Thus, a rapid response by experienced staff, with specific equipment, close monitoring, and multidisciplinary management are essential.

Fig. 1
figure 1

Therapeutic plasma exchange: overview. TPE therapeutic plasma exchange, HIV human immunodeficiency virus

The goal of this article is to present a narrative review of the main indications for TPE in critically ill patients, as well as their main characteristics. A multidisciplinary group of intensivists, immunologists, nephrologists, pathologists, and hematologists reviewed and summarized the evidence on the rationale and indications for TPE in the ICU, shared their experience, and identified relevant issues that need to be known by the intensivists, as well as knowledge gaps that need to be filled by future research.

Indications for urgent TPE in critically ill patients

The American Society for Apheresis (ASFA) updated its guidelines on therapeutic apheresis in 2019 [2], and the Japanese Society in 2021 [3]. They identified four categories of use: first-line therapy (Category I), second-line therapy (Category II), role not established (Category III), and ineffective or harmful (Category IV). In the ICU, the indications for TPE can be divided into (1) absolute, well-established, and evidence-based, for which TPE is recognized as first-line therapy, (2) relative, for which TPE is a recognized second-line treatment alone or combined with other interventions and (3) rescue therapy, where TPE is used with limited evidence of benefits but a plausible theoretical rationale (Table 1) [4,5,6,7].

Table 1 Indications for therapeutic plasma exchange (TPE) in the ICU: absolute (likely or less likely be used), relative, and rescue therapy

Mechanisms, kinetics, and goals of TPE

Mechanisms of TPE

TPE has two mechanisms of action (Fig. 1):

  1. 1.

    Removal of a pathogenic substance from the plasma (e.g., IgG in myasthenia gravis, IgM in Waldenström macroglobulinemia, or IgG and IgM iso-agglutinins prior to ABO incompatible organ transplantation [8]). To be efficiently cleared by TPE, the substance should ideally be identified and assayed and have a high molecular weight, low distribution volume (chiefly in plasma), long half-life, and low turnover rate. Of note, the degree of substance removal does not necessarily correlate with the alleviation of the clinical symptoms like in myasthenia gravis [9].

  2. 2.

    Delivery of large amounts of deficient plasma components (e.g., ADAMTS13 in thrombotic thrombocytopenic purpura (TTP)). The fluid used for plasma replacement should be, or be derived from, healthy donor plasma [1].

Kinetic models

Kinetic models for prediction of substance removal have been developed [10]. The half-life and volume of distribution of the substance to be removed must be considered when planning the intensity and frequency of TPE sessions. The plasma volume to be replaced is determined by calculating the total blood volume and the total plasma volume (TPV) of the patient [11]. For a substance that is neither rapidly synthesized nor redistributed and limited to the intravascular space, the first session of plasma exchange will remove 65–70% of the target substance. With additional plasma volumes exchanged, the absolute amount removed becomes progressively smaller due to the exponential nature of the removal (Fig. 2) The second session will remove an additional 23% and the third session only an additional 9% of the target substance. The net reduction will be affected by the redistribution from extravascular to intravascular compartments, production rate and by volumes of distribution. For example, one standard TPE session replacing 1.2 times the TPV will remove 10 g of IgG and 0.3 g of IgM due to the amount of IgG present in the intravascular space and its ability to redistribute from the extravascular compartment, which does not occur in an appreciable amount with IgM [12]. It also depends on the level of IgG at baseline (Fig. 2). In patients who are IgG depleted, TPE can replace the missing IgG [13].

Fig. 2
figure 2

Progressive decrease in plasma concentration of substance following four consecutive TPE treatments equaling 1.2 plasma volume each. TPE therapeutic plasma exchange

The 2019 ASFA recommendations suggest exchanging 1.0–1.5 times the individually calculated TPV [2]. However, several clinical studies have shown a frequent failure to reach this TPE target [14]. A study in Germany reports exchanging only 0.4–1.0 times the estimated TPV [15]. In a recent study from India, the overall exchange volume during TPE for various indications was only 2.1 L with an overall response rate of 84% [16]. The optimal exchange volume is not known and may depend on the disease. Small volume plasma exchange will remove less substances from the plasma but may be more affordable and still effective. For instance, in Bangladesh, where most patients with Guillain–Barré syndrome (GBS) cannot afford standard treatment with intravenous immunoglobulin or a standard TPE course, a small clinical study in 20 adult patients with GBS demonstrated the feasibility and safety of small volume plasma exchange as a potential alternative low-cost treatment [17]. A detrimental effect of high-dose TPE has not been described but it should be remembered that TPE also removes drugs that are aimed at treating the underlying disease, such as rituximab or caplacizumab or essential drugs such as antibiotics or anticoagulants. Also, if the aim is to remove larger substances, the efficacy of TPE will decrease as the total exchanged volume increases, as the removed larger amounts of a pathologic substance may need hours to days to diffuse from the extravascular to the intravascular compartment [12]. In this case, it may be more efficacious to repeat TPE sessions rather than continuing high-volume TPE beyond 1–1.5 plasma volumes. Knowledge about the characteristics and kinetics of the substance(s) to be removed is essential to guide the TPE prescription. The most rational approach to achieve the most efficient substance removal is to consider the nature of the toxin(s) to be removed and the best combination of exchange volume, treatment frequency and timing [18].

Therapeutic goals of TPE

The therapeutic goals of TPE depend on the pathophysiology of the disease. For instance, in Waldenström macroglobulinemia, the goal is to decrease the IgM level to reduce plasma viscosity and eliminate symptoms of hypoperfusion. In TTP, the aim is to raise the platelet count above 150,000/µL and reversing hemolysis by removing anti-ADAMTS13 inhibitory antibodies, removing ultralarge von Willebrand factors multimers and replacing ADAMTS13 enzyme [19]. In myasthenia gravis, the aim is to achieve a rapid clinical stabilization by removing acetylcholine receptor antibodies, especially in case of myasthenic crisis. In GBS, the goal is to improve muscle strength and to reduce the need for mechanical ventilation and hasten recovery. Table 1 shows the main parameters to monitor and endpoints for the different TPE indications in the ICU (Table 1).

Diagnostic workup for TPE indications and monitoring

TPE is used in various medical conditions. The diagnostic work-up serves to identify the underlying disease and determine its characteristics (Table 2). During TPE, close monitoring is essential to prevent adverse events and to ensure efficacy and safety. The criteria for discontinuing TPE should be determined a priori. Many routine biomarkers (e.g., C-reactive protein (CRP), creatinine, bilirubin etc.) will be reduced after a TPE session, potentially for many hours, and therefore, must be interpreted with caution. Changes in the amount of a substance removed by TPE may not necessarily represent improvement in the disease process and additional evidence of clinical response such as symptom resolution should be sought (Table 1S). Similarly, a decrease in CRP level after TPE does not necessarily mean that inflammation and/or infection are under control.

Table 2 Disease-specific workup for the most common indications

Technical aspects

Machines and devices

During TPE, the plasma can be separated from the corpuscular components of the blood by centrifugation, membrane filtration, or both [20]. Centrifugation is based on the differences in density of the various blood components. Mature red blood cells (RBCs) have the greatest relative density, followed by young erythrocytes (neocytes), granulocytes, mononuclear cells, platelets and, finally, plasma. Filtration takes advantage of differences in particle size to separate plasma from cells.

Currently licensed TPE devices can operate with a continuous or an intermittent flow [21]. Both, centrifugal and membrane-based devices are available. In apheresis units based in transfusion medicine or hematology departments, TPE is usually performed with centrifugal systems (cTPE) that often use citrate for anticoagulation. In most nephrology departments and ICUs, the preferred devices are membrane-based (mTPE), including multifunctional renal replacement therapy (RRT) machines. In both cTPE and mTPE, the cell-rich blood that remains after plasma removal is mixed with the replacement fluid (e.g., albumin, plasma, or crystalloid) and returns to the patient to prevent hypovolemia. To reduce costs and donor exposures, up to 30% of the replacement fluid may be a suitable crystalloid. In low-resource healthcare systems, plasma, crystalloid, or non-plasma colloid beyond 30% of the replaced volume may be used for replacement due to the expense of albumin substrates, and availability and safety profile of plasma products.

Plasma removal efficiency (PRE) is the metric used to compare TPE devices. It describes the fraction (%) of plasma that passes through the device and is removed per procedure. PRE estimate may vary according to the mathematical formulas used [22,23,24,25,26]. With cTPE devices, PRE is faster and higher than with mTPE devices [12, 26]. Rates of removal are comparable with cTPE and mTPE for IgG but not for fibrinogen [12].

Vascular access

The choice of vascular access for TPE depends primarily on the method used: cTPE typically requires lower blood flow rates (Qb) (50–120 mL/min) than mTPE (150–200 mL/min) [27]. A lower Qb enables the use of narrower catheters such as peripheral devices (e.g., 18-Gauge needle) or standard triple-lumen central venous catheters (e.g., 7 Fr). With a peripheral vein, single-needle access is feasible when using cTPE [28] but might increase the treatment time. Peripherally inserted central catheters are not suitable because their narrow catheter gauge will collapse with the negative pressures exerted during TPE [29]. The mTPE devices often require higher Qb and, therefore, wider catheters such as temporary hemodialysis catheters or large-diameter dual-lumen catheters (e.g., 13.5 French) [30]. The optimum characteristics of a catheter for TPE include rigid walls, a large diameter, and a short length to reduce resistance and decrease instrument alarms. Machines used for cTPE can concentrate RBCs to a hematocrit of 80% or higher, which allows for more plasma per volume to be processed compared to mTPE devices [11]. A higher Qb is needed with mTPE devices as they usually extract only about 30–35% of processed plasma to prevent RBC damage from a high hematocrit. Thus, with mTPE devices three or four times more plasma volume must be processed to remove similar plasma volume as with cTPE devices.

Anticoagulation

Anticoagulation for TPE aims to achieve a delicate balance between preventing circuit failure with loss of expensive blood components and preventing bleeding. Systemic heparin and regional citrate are the most common anticoagulants, while epoprostenol can also be used, when citrate is unavailable, and heparin is contraindicated. In the past, citrate was generally used for cTPE and heparin for mTPE, but citrate is now also used for mTPE [12, 31, 32]. According to the World Apheresis Registry, in which two-thirds of apheresis procedures were therapeutic, 73% of procedures were provided with citrate anticoagulation [33].

Both heparin and citrate anticoagulation have advantages and drawbacks (Table 2S). The risk of bleeding during TPE is lower with citrate than with heparin. However, when citrate is used with a mTPE device, side effects are more frequent, mainly because more citrate is required as a result of a higher Qb, plus, removal of less plasma leads to more citrate entering the patient’s systemic circulation [11]. Symptomatic hypocalcemia is also more common with citrate and can be prevented by prophylactic calcium administration [34]. Commercially available mTPE devices with integrated citrate administration adjusted for Qb and calcium supplementation according to effluent rate reduce the risk. When using heparin for anticoagulation, estimation of the required dosage should factor in extracorporeal losses of the drug and its cofactor antithrombin [35]. Moreover, antithrombin loss may hamper anticoagulation with heparin as well as the interpretation of chromogenic anti-Xa assays that add exogenous antithrombin.

Fluid replacement

Albumin or plasma can be used as replacement fluid, alone or in combination, and with or without the addition of a crystalloid such as saline. Albumin is used most often, as it is associated with a lower frequency of allergic or immune reactions (e.g., transfusion-related acute lung injury) compared to plasma and not associated with a risk of transfusion transmitted disease [12, 36, 37]. Table 3S summarizes pros and cons of each alternative (Table 3S). When albumin is used as replacement solution, metabolic acidosis may be seen after the TPE session because albumin has an acidic profile [38]. Albumin substitution may also affect the concentrations of fibrinogen and other coagulant factors resulting in profound derangement of thromboelastography parameters [39].

Plasma is indicated when aiming to replace plasma components (e.g., ADAMTS13 in TTP). Despite the absence of hard evidence, many centers also use plasma to prevent depletion of coagulation factors (e.g., if a bleeding diathesis is present or an invasive procedure is planned). Established guidelines for hemostasis monitoring/management during TPE are lacking but the extracorporeal losses of both pro- and anticoagulant factors need to be considered [40].

A recent survey by an ASFA subcommittee found wide practice variation in the type of replacement fluid but the potential bleeding risk most often determines the choice [41]. Because of the large volume, the number of donor exposures, and often prolonged duration of therapy, the risk of allergic reactions is higher with plasma than with albumin, and some centers administer antihistamines and/or glucocorticoids when using plasma [42]. When plasma is used as replacement solution, metabolic alkalosis may occur because of metabolism of citrate used as anticoagulant and citrate present in stored plasma. For every citrate molecule metabolized, there is a consumption of hydrogen ions and production of three sodium bicarbonate molecules, thus increasing serum pH levels [43].

Crystalloid can be added for cost-containment and in patients with hyperviscosity syndrome. However, replacing plasma with crystalloid carries a risk of hypotension if the proportion of replacement with crystalloid exceeds 30% [44]. In this setting, significant fluid shifts can occur as water follows its concentration gradient from the intravascular space into the extravascular space. When crystalloid is used as a portion of the replacement, it should be administered at the beginning of the exchange and not at the end to avoid significant fluid shifts and hypotension. Hydroxyethyl starch (HES) is no longer recommended in critically ill patients due to its harmful effects on both renal function and coagulation. However, it is still occasionally used as a replacement fluid (e.g., 3% HES with 5% human albumin), especially in low-resource healthcare systems [45, 46]. It may also be used in patients who refuse blood products.

Clinical response

The expected benefits and potentially deleterious effects of TPE are dependent on the timing of the procedure with respect to the onset of the illness, the volume of fluid exchanged, the type of replacement solution, and the frequency and intervals of plasma removal. The individual criteria for “clinical response” are highly disease specific, ranging from changes in individual or multiple hematological parameters, antibody concentrations or biochemistry to improvement of clinical signs and symptoms. The impact of TPE can be rapid or slow and may last for weeks to months, depending on the underlying disease. However, long-term effects, including psychological well-being and the risk of chronic organ dysfunction beyond the acute illness are rarely reported.

Complications

TPE is a relatively safe procedure and usually well tolerated. Complications include catheter-related and procedure-related events. The incidence of adverse events has declined over time [47, 48] and now ranges from 5 to 36% depending on vascular access used, type of replacement fluid, and anticoagulation (Table 4S). Catheter-related infections, pneumothorax, and local bleeding have been reported in 0.4–1.6% of patients [49, 50]. In critically ill patients, bleeding disorders were rare (< 10%) but catheter dysfunction was the most common complication (32%) [30]. Complication rates were similar with mTPE and cTPE [30]. Potentially life-threatening complications, dominated by anaphylactoid reactions and severe hypotension, have been reported in 1–2% of TPE sessions in critically ill patients [30, 51]. They should be minimized by the judicious choice of a vascular access in close collaboration with the apheresis specialist.

Citrate anticoagulation and plasma replacement are risk factors for hypocalcemia and paresthesia [52]. Plasma replacement is associated with a higher risk of anaphylactoid reactions. On the other hand, replacement with albumin does not correct the depletion and balancing of coagulation factors and immunoglobulins, resulting in a potential risk of bleeding and infection, respectively.

Drug removal by TPE

Data on drug removal by TPE are scarce and based on case reports or case series only [53, 54]. For most drugs, either no information is available, or it is not important. For highly protein-bound drugs with a low volume of distribution, and for chimeric antibodies, there is very effective removal. Factors associated with clinically meaningful drug removal include drug characteristics (volume of distribution, protein-binding affinity, rate of endogenous clearance, distribution half-life, dose-related pharmacodynamics), TPE characteristics (volume of plasma removed, interval between sessions, time between first and last session) and timing of drug administration [54,55,56,57]. Important inter- and intra-individual differences in pharmacokinetics and the multi-compartmental kinetic patterns seen during TPE can make predictions very difficult.

Antibiotic removal during TPE was reviewed recently [53, 56]. Whether an antibiotic should be administered before or after TPE depends also on its pharmacodynamic characteristics. Aminoglycosides can be best administered before the procedure to benefit from both a high peak with bactericidal effect and reduced toxicity related to a low trough level through extracorporeal removal. Beta-lactam plasma levels, on the other hand, should be maintained above the minimum inhibitory concentration which often requires a supplementary dose post-procedure. Monoclonal antibodies such as rituximab have a small volume of distribution and a long distribution half-life and therefore are significantly removed by TPE [58]. During TPE, total clearance of the drug decreases over time as the plasma levels decrease [59]. Although levels of monoclonal antibodies correlate with clinical effects, they may not correlate with pharmacodynamic markers (i.e., the CD20 + B-cell count for rituximab) [54]. Significant removal of enoxaparin, tacrolimus, and mycophenolic acid during TPE has been reported [60, 61]. Most studies involved administering medications after TPE and scheduling the next TPE session 24–36 h later. In general, therapeutic drug monitoring should be applied whenever possible in critically ill patients undergoing serial TPE sessions, especially if the drug has a narrow therapeutic index. Timing of sampling should account for post-procedure redistribution with rebound of plasma concentration.

Unanswered questions and research agenda

Potential novel mechanisms and emerging ICU indications for TPE

For the most urgent TPE indications in critical care listed in Table 1, the efficacy of TPE is thought to stem from the removal of pathogenic substances and/or provision of deficient protective molecules. This classical blood purification concept may apply to systemic inflammatory syndromes encountered in a wide variety of critical conditions, but timing and anti-/pro-inflammatory balance may be pivotal in determining benefit versus potential detriment. Thus, inflammatory processes with consumptive coagulopathy ranging from thrombocytopenia to disseminated intravascular coagulation might respond to TPE. Furthermore, TPE removes damage-associated molecular patterns (DAMPs) that are released by injured cells and may trigger and perpetuate multiorgan dysfunction. In patients with sepsis and multiorgan dysfunction, TPE can lead to shock reversal and improve vascular permeability and coagulation abnormalities, while also producing a trend toward improved survival [62,63,64].

Given the ability of TPE to modulate systemic inflammation and coagulopathy, potential benefits in patients with severe COVID-19 have generated interest [65, 66]. Moreover, TPE can correct the increased von Willebrand factor multimer and the decreased ADAMTS13 activity in COVID-19 patients [67]. Faster recovery but no effect on mortality was shown in one small randomized controlled trial [68]. Many studies, including randomized controlled trials, are ongoing to test various hypotheses using slightly different protocols. Apart from sepsis, clinical scenarios characterized by a systemic inflammatory response that may improve with TPE include hemophagocytic lymphohistiocytosis, macrophage activation syndrome, chimeric antigen receptor T-cell-associated cytokine release syndrome, severe pancreatitis, and severe burns. So far, the current evidence remains limited to case series and uncontrolled observational studies. Finally, TPE has been used recently in refractory cases of vaccine-induced thrombosis and thrombocytopenia which could be added to the list of rescue therapy although evidence is still limited [69].

Initiation of TPE

The appropriate timing of TPE initiation needs to be determined. Biomarker levels, antibody titers, or clinical symptoms that support TPE initiation vary across indications. Specific cut-offs associated with poor outcomes need to be identified. Of note, the inflammatory syndromes encountered in the ICU may also serve as markers for monitoring of the effectiveness of TPE, such as markers of endothelial activation and primary hemostasis. Although trauma and sepsis are different entities, in both, elevations of glycocalyx-shedding biomarkers such as syndecan-1 and heparan sulfate are associated with poor outcomes [70] and their levels can be reduced with TPE [71]. Also, an imbalance between ADAMTS13 and von Willebrand factor is found in both sepsis and trauma. Specific cut-offs have been suggested, but whether these are useful to guide TPE remains unknown.

Comparison of TPE to other interventions

For most conditions, the efficacy of TPE compared to other techniques is not known. In GBS and myasthenia gravis, the effectiveness of TPE was compared to that of IVIG or a combination of both [72]. For conditions related to a pathogenic antibody, limited-level evidence suggests that TPE and more selective immunoadsorption techniques might have similar efficacy, but more studies are needed. Also, new data may challenge the benefit of TPE in some instances. Trials such as the PEXIVAS study led the AFSA to change severe ANCA-associated vasculitis from a category I to category II indication for TPE [73, 74].

Technical aspects of TPE

Little evidence supports the standard TPE regimens in ICU patients. More specifically, all current regimens were developed based on long-term experience with ward patients or outpatients. ICU patients likely have altered volumes of distribution due to organ failures, capillary leakage, and/or hypoalbuminemia. Ideally, TPE regimens should be tailored to the needs of the individual patient. More information about the optimal TPE intervals and volumes for critically ill patients is needed, as well as the optimal replacement solutions and the stopping cut-offs associated with a low risk of rebound.

Conclusions

TPE is an established therapy in modern critical care. It includes centrifugal and membrane-based techniques and requires fluid replacement with plasma or albumin solution. We have summarized the key points for the non-TPE specialists (Table 3). Although TPE is considered as first- or second-line therapy in many disorders, significant knowledge gaps remain, especially with regard to the exact triggers and cut-offs for initiation, optimal markers for monitoring and triggers for discontinuation. Furthermore, the interpretation of routine laboratory blood tests and drug dosing are challenging during TPE. More observational and interventional studies are needed to fill the existing knowledge gaps, to identify patients who are likely to benefit from TPE and to avoid TPE in those who will not benefit or may come to harm.

Table 3 Key points for the non-TPE specialists