Plasmapheresis for neuroinflammatory disorders


Ralf Gold MD, Department of Neurology, St. Josef-Hospital, Ruhr-University, Bochum, Gudrunstrasse 56, 44791 Bochum, Germany.
Tel: +49-234-5092411
Fax: +49-234-5092414


Nowadays, therapeutic plasma exchange (PE) is the most common apheresis procedure. Plasma is separated from corpuscular blood constitutents and replaced with a substitution fluid. Thus, PE is a non-specific treatment modality. Its therapeutic effect is based on the removal of circulating, pathogenic immune factors, including autoantibodies. To date, PE is well established for several immune mediated neurological disorders. Whilst the first experience was acquired in acute life-threatening conditions, such as the treatment of Guillain–Barré syndrome or myasthenic crisis, therapeutic success was later achieved in other chronic autoimmune diseases; PE was applied successfully in chronic inflammatory demyelinating polyneuropathy, paraproteinaemic polyneuropathy, stiff-person syndrome and might also be tried in autoimmune diseases of paraneoplastic origin. From a novel aspect, PE was also established as an escalation therapy for steroid unresponsive relapses of multiple sclerosis, and thus has gained even more widespread attention. Humoral disease mechanisms dominate even more in neuromyelitis optica, as a subtype of MS, and usually respond to PE. Adding to its increasing application in clinical practice, the procedure is usually well tolerated. Possible adverse reactions of PE arise from the large-size vascular access site, the use of replacement fluids and the need for anticoagulation. (Clin. Exp. Neuroimmunol. doi: 10.1111/j.1759-1961.2010.0010.x, 2010)


Plasmapheresis is a non-selective immune procedure that separates the entire plasma from the corpuscular blood components. This can be achieved by centrifugation or membrane filtration. Although plasmapheresis and plasma exchange differ slightly in concept, today both terms are often used synonymously in the literature.

Plasma exchange procedures were first introduced into human therapy for the treatment of hyperviscosity syndrome. Currently, it has been well established to manage a number of acute neurological, hematological and other autoimmune disorders, including myasthenic crisis, Guillain–Barré syndrome, sickle-cell disease or Goodpasture’s syndrome (for overview see Table 1) and is also under debate for other conditions, such as acute hearing loss or macular degeneration, where the evidence for efficacy is less clear.1 As early as the late 1970s, the efficacy of PE was shown for the treatment of myasthenic crisis2 or Guillain–Barré syndrome (GBS).3 Already in the late 1970s, PE was used for unselected treatment of severe disease courses of multiple sclerosis (MS).4 At this time, its use in chronic MS had not been successful and had been abandoned for two decades, until novel molecular histopathological data reversed pathophysiological concepts.5 To date, therapeutic PE has been established as escalation therapy of steroid-unresponsive MS relapses.6–8 Moreover, further neurological indications have been defined, including escalation therapy for neuromyelitis optica (NMO, Devic’s syndrome)9 and diseases of the peripheral nervous system with severe or life-threatening symptoms.3

Table 1.   Evidence-based use of therapeutic apheresis as categorized by the American Society for Apheresis
In clinical practice
 Acute exacerbation of myasthenia gravis or preoperative stabilisation
 Guillain–Barré syndrome for patients with severe impairment (grade  3–5)
 Chronic inflammatory demyelinating polyneuropathy with no   response to conventional therapy
 Goodpasture syndrome
 Refsum’s disease
 Thrombotic thrombocytopenic purpura
 Post-transfusion purpura
Generally accepted indications
 Acute fulminant demyelination of the central nervous system
 Idiopathic thrombocytopenic purpura
 Hyperviscosity syndrome/paraproteinemia
 Lambert–Eaton syndrome
 AB0 mismatched bone marrow transplantation
 Rapid progressive glomerulonephritis
 Demyelinating polyneuropathy (IgM) in Waldenström’s  macroglobulinemia

The therapeutic effect of apheresis procedures in autoimmune diseases is mainly based on a modulation of the humoral immune system. It typically involves the elimination of pathogenic autoantibodies, complement and cytokines. Removal of humoral factors might additionally modulate cellular components of the immune system. Linkage especially occurs through cell types with receptors for immunoglobulins (Fc-receptors), such as monocytes, macrophages and natural killer cells, which are a focus of interest. Yet, such effects on cellular immune responses are so far only incompletely understood. Recent findings show that extracorporeal removal of immunoglobulins by adsorption techniques might also impact on T cell mediated immune responses in addition to the immunoglobulin reducing effect.10,11

Yet, the modulatory effect of plasma exchange on the immune system does not persist over time. Hence, the frequency of apheresis treatment needs to be determined individually according to the underlying disease, the patient’s improvement and the rebound of pathogenic factors. Based on theoretical considerations and an exponential elimination kinetics of humoral plasma factors, a maximum of five sessions is usually carried out per treatment cycle.

Plasmapheresis versus immunoadsorption techniques

Apheresis procedures generally require one line for blood withdrawal and another line for reinfusion; this can also be carried out through a double-lumen Shaldon catheter. After protein separation, oncotic pressure in the plasma is restored by adding human albumin. Historically, fresh frozen plasma was given in the 1980s. During the exchange procedure, anticoagulation with heparin or citrate is needed to prevent clotting and thrombus formation.

Plasmapheresis is non-selective and thus 150 g of plasma protein has to be removed to eliminate approximately 1 g of pathogenic antibodies. Rare side-effects include hemodynamic instability, dilutional coagulopathy, hypocalcemia and allergic reactions.12 Furthermore, catheter related complications might involve thrombosis, septic infections or pneumothorax.13 Medical costs are mainly restricted to protein replacement with human albumin, which, as a blood product, harbours a minor risk of infection in industrialized countries.

In contrast to entire plasma elimination with plasmapheresis, immunoadsorption is clearly a more selective technique for removing IgG antibodies by binding to a specific matrix (e.g. protein A or tryptophan). Immunoadsorption devices can be subdivided into non-selective, semi-selective and highly selective adsorbers with different affinities to plasma proteins. Non-selective adsorbers, such as dextran-sulphate, tryptophan and phenylalanine, reduce the plasma levels of several different substances, such as fibrinogen, albumin, lipids and immunoglobulins. Semi-selective protein A columns are treatment devices that contain highly-purified staphylococcal A protein bound to an inert silica matrix. Separated plasma is passed over adsorbant columns containing protein A as a specific ligand that binds IgG antibodies. The processed plasma is given back to the patient while anticoagulation is used during the process. Thus, plasma product replacement is not necessary. Technically, there are single-use and re-usable adsorbers available. Although serious side-effects are rare, leukocytoclastic vasculitis following staphylococcal protein A column has been described after immunoadsorption therapy.14 Immunoadsorption techniques were first described in autoimmune myasthenia as prototypical antibody-mediated disease.15,16

Plasma exchange in neuromyelitis optica

Neuromyelitis optica (NMO), or Devic’s disease, is an increasingly better understood demyelinating disease of the central nervous system (CNS). It predominantly involves the optic nerve and the spinal cord, where the white as well as gray matter can be affected. Additionally, intracranial lesions might be observed in some cases. In NMO, lesions show a prominent perivascular distribution. They are characterized by complement and antibody deposition, as well as eosinophil infiltration. In some instances, tissue necrosis and severe disability can also occur. In recent years, the pathogenic role of astrocytes in the pathophysiology of the disease has been increasingly recognized. In 2004, a specific antibody called NMO-IgG was first described as a new biomarker associated with NMO.17 Later, NMO-IgG was characterized as an antibody against aquaporin-4, a water channel which is expressed on processes of perivascular astroglia.18 Recently, the passive transfer of anti-aquaporin-4-antibodies from NMO patients in animal models also proved its pathogenetic relevance to this disease.19–21

Traditionally, high dose glucocorticosteroids (GS) constituted the first-line treatment for NMO. Yet, in steroid unresponsive NMO relapses, PE evolved as a well-working escalation therapy. In 2002, the efficacy of PE as a rescue therapy in NMO relapses was first shown in an open-label study initiated by the Mayo Clinic.8 These data were further corroborated by Watanabe et al. They reported on a series of six NMO-IgG-positive patients, three with acute optic neuritis and three with myelitic relapses, who did not improve with high-dose GS treatment.9 In response to PE, three of these patients showed a definite functional improvement as early as after the first or second exchange.

Plasma exchange in multiple sclerosis relapses

The molecular groundwork for plasmapheresis was laid out by histopathological studies identifying the local deposition of antibodies and complement in acute demyelinating lesions.5 Based on a large series of diagnostic brain biopsies from MS patients, this international group of investigators coined a four pattern classification of actively demyelinating MS lesions based on immunopathological characteristics with distinct mechanisms of demyelination.22 In this setting, the most relevant aspect of plasma exchange is the “type II” pattern, with local deposition of antibodies and complement pointing to a pathologically relevant role of B cells and plasma cells. The value of GS is certainly limited in this scenario. Although GS might downregulate cellular cytotoxicity and lead to the death of activated B cells,23 they will not lead to immediate effects on conduction blockade by local antibody deposition.

The knowledge of this recent concept of immunopathologically distinct MS subtypes makes a strong point for individualized therapeutic approaches. The direct elimination of humoral factors through PE therapy has shown beneficial effects in severe GS-refractory MS relapses; first, PE was tested in a randomized, sham controlled and double blind crossover trial by Weinshenker et al. including patients with an acute, severe neurological deficit caused by MS or other demyelinating diseases of the CNS.8 All patients did not respond to high-dose GS pulse within the last 3 months of treatment. In this situation, PE led to an improvement in 42%, as compared with 6% after sham treatment. Importantly, further subanalyses showed that initiation of treatment within 6 weeks after the onset of symptoms was predictive for therapeutic outcome.24 To support this concept, Keegan et al. correlated the treatment response to PE with the (blinded) immunopathological pattern of lesions in patients who underwent a brain biopsy and also received PE.25 On retrospective analysis of 19 consecutive patients, only those 10 with a “type II” pattern in the brain biopsy showed at least a moderate neurological improvement. In contrast, patients with other MS subtypes did not respond to PE. In 2004, the value of PE could be extended to patients suffering from MS or a clinically isolated syndrome with severe corticosteroid-refractory optic neuritis.6 Here, improvement of visual acuity was observed in seven of 10 patients receiving GS pulse therapy in a similar pattern. On follow up, three of the 10 patients continued to improve, two remained stable and two worsened again. Treatment success was better when the residual visual acuity was greater than 0.05 (20/400), showing that in severe cases, edema might lead to additional secondary ischemic optic neuropathy not responsive to PE. Indeed, the efficacy of PE might be a result of the elimination of humoral factors that lead to conduction block with clinical impairment before severe structural damage has occurred.

In the meantime, several uncontrolled case series additionally confirmed the sustained efficacy of PE in patients with steroid refractory MS relapses.6–8 All these studies show a treatment response of approximately 70%, with a median onset of improvement determined after the third session of therapeutic PE.7 Importantly, less than a 6-week interval should occur between the onset of relapse and the initiation of PE series. To date, it is still unclear whether intravenous immunoglobulin (IVIG) can substitute for PE as an escalation treatment in MS relapse. Although IVIG did not show sustained beneficial effects as a disease modifying treatment in MS, a recent retrospective study showed some benefit of IVIG for the improvement of visual acuity in steroid unresponsive optic neuritis.26 No published series exists so far about the efficacy of immunoadsorption in the setting of steroid-non-responsive MS relapses.

Although PE therapy in primary and secondary chronic progressive MS is not effective and should not be carried out, efficacy of PE for superimposed relapses in secondary progressive MS was reported in two cases.27,28 Improvement of patients’ mobility with an amelioration of quality of life was described. Thus, the decision for PE might be made on an individual basis and in some selected cases can be guided by the identification of the local deposit of humoral factors in a brain biopsy.27

Plasma exchange in other inflammatory diseases of the central nervous system

Besides comprehensive data on the efficacy of PE in MS, evidence for the efficacy of plasma exchange in other inflammatory diseases of the CNS, such as Bickerstaff brainstem encephalitis, has only been provided in uncontrolled case reports so far.29,30 These data include rare diseases, such as autoimmune and paraneoplastic encephalitis, as well as CNS manifestations of systemic vasculitis. Here, the efficacy of PE has been shown for CNS complications of systemic lupus erythematosus.31 Similarly, PE might be successfully applied for treating neurological complications of other systemic rheumatological and also hematological diseases. For example, PE has been used to treat thrombocytopenic purpura since the 1980s and its efficacy has recently been corroborated in a retrospective analysis of 17 patients.32,33 Of note, the efficacy of PE is only seen during acute deterioration and has to be followed by the use of long-term immunotherapy.

Although controlled data are limited, PE might be similarly effective in classic paraneoplastic syndromes of the CNS, such as limbic encephalitis or cerebellar degeneration.29,30 As another autoimmune and, sometimes, even paraneoplastic disease, the stiff-person syndrome (SPS) is characterized by central hyperexcitability and cramping or muscle spasms. In some, but not all cases, antibodies against amphiphysin or glutamate decarboxylase can be found. The relevance of such humoral factors in SPS has recently been confirmed by passive transfer of serum from patients to animal models.34 Some patients suffering from SPS can be well managed with GS. In other cases, a benefit from PE was described in several uncontrolled reports.35

In Japanese patients with myelitis and atopic diathesis, a retrospective study analyzed the therapeutic efficacy of PE versus IVIG application in a four-arm setting.36 Although slightly limited by low patient numbers, this study suggests an enhanced efficacy in both groups, including PE, to the therapeutic regimen. Finally, the successful application of PE was also reported in single cases of Weston–Hurst acute hemorrhagic leukencephalitis.37,38

PE for diseases of the neuromuscular endplate

Autoimmune diseases of the neuromuscular endplate represent classic paradigms for antibody mediated disorders and thus represent ideal targets for PE therapy. In the 1970s, transfer studies in animal models provided elegant evidence for the pathogenic relevance of antibodies against acetylcholine receptors in myasthenia gravis. Consequently, myasthenia gravis was one of the first neurological disorders where PE was implemented. Although the first studies in myasthenia were carried out as early as in the 1980s,2,39 the efficacy of PE in the related Lambert–Eaton myasthenic syndrome (LEMS) has so far only been described in case series.40

Until now, several controlled trials and meta-analyses have provided evidence for the efficacy of PE in myasthenia gravis. The first randomized trial included a total of 87 patients with myasthenic crisis or severe impairment with high-grade muscular weakness. Patients were randomized to PE or, alternatively, 0.4 g/kg IVIG on three or five consecutive days.41 After 15 days of treatment, there was an improvement in muscle strength in all three groups as compared with baseline, but no superior efficacy of any regimen. Another retrospective, multicenter study included 54 patients with myasthenic crisis and, again, compared the efficacy of PE versus IVIG.42 This trial pointed to a better tolerability and a shorter time of hospitalization for patients treated with IVIG. Yet, the need of artificial ventilation after 14 days was lower in patients receiving PE. A most recent study addressed the question of combination therapy with PE and immunoadsorption in patients with myasthenic crisis.43 In a retrospective analysis covering 80 patients with myasthenic crisis, regimens including immunoadsorption led to a shorter hospital stay and a better functional outcome. In addition, patients treated with a combination of PE and immunoadsorption suffered from fewer side-effects.44 In summary, these data suggest that the combination of PE and immunoadsorption might serve as optimized therapeutic management in the situation of myasthenic crisis.

Finally, neuromyotonia (Isaacs–Mertens syndrome) is a rare, immune-mediated disorder of the neuromuscular endplate characterized by cramping, myokymia, fasciculations and excessive sweating. The frequent presence of antibodies against voltage-gated potassium channels in the serum provides a rationale for the successful application of plasmapheresis. Yet, the successful application of PE in neuromyotonia has so far only been described in case reports or small case series.45 Some studies describe the elimination of autoantibodies or the normalization of peripheral nerve hyperexcitability after PE.46 Another group reported on a patient who was not responsive to IVIG, but finally improved after treatment with PE.45

Plasma exchange in neuroimmunological disorders of the peripheral nerve

The hallmark indication for the application of PE in diseases of the peripheral nervous system is the Guillain–Barré–Strohl Syndrome (GBS). GBS is an immune-mediated, demyelinating acute neuropathy that presents with initial sensory disturbances and subsequent ascending flaccid paralysis, often after a gastrointestinal or upper respiratory infection. GBS might lead to severe sequelae, such as autonomic failure, respiratory insufficiency, permanent disability and even death, in a significant number of cases. The efficacy of PE in GBS was reported as early as in 1985.3,47,48 In a subsequent randomized trial that included a total of 245 patients, PE did not only significantly improve the short-term motor function after 4 weeks, but also influenced the long-term outcome at 6 months.49

In this study, PE showed a particular efficacy when applied early in the course of the disease and in the subgroup of patients with respiratory failure. A more recent and large trial in GBS patients compared the efficacy of PE versus IVIG and the combination of both regimens. This study showed an equivalent efficacy of PE and IVIG monotherapy. However, the sequential application of both regimens did not lead to any additional benefit.47,50 While these studies were carried out in adult patients, there are only few data available in children or in GBS subtypes, such as the Miller–Fisher variant.51,52 A German group reported on the successful application of immunoadsorption therapy in an 11-year-old boy with GBS and severe motor involvement, as well as respiratory insufficiency. Recently, a retrospective study analyzed the efficacy of IVIG or PE for treatment of patients suffering from Miller–Fisher syndrome. In this study, neither IVIG nor PE was superior to supportive therapy alone. While the reason for this observation is not definitely clear, it might be due to the benign, self-limiting course of the disease in most cases.52

Chronic inflammatory demyelinating polyneuropathy (CIDP) is a disease closely related to GBS. Yet, CIDP is rather characterized by a relapsing or slowly progressing disease course, whereas GBS is usually acute and monophasic in nature. Sural nerve biopsy in CIDP shows a chronic neuropathy with inflammation and patterns of de- and remyelination. The efficacy of PE in CIDP was only tested in two smaller, but controlled trials.53,54 One study included 18 participants who received 10 exchanges over 4 weeks in a crossover design. Here, PE led to a significant improvement of disability in comparison to sham exchange. These data were corroborated in a meta-analysis, as well as in another parallel group trial with 29 participants.55 Yet, in the cross-over study, rapid re-deterioration occurred in 75% of PE responsive patients after the end of therapy. These data nicely underpin the transient effects of PE and the need for long-term follow-up therapies after successful PE in chronic disorders. They do not explain how putative responders towards PE versus intravenous immunoglobulin can be predicted for an individualized treatment paradigm.

In multifocal motor neuropathy (MMN), patients suffer from an acquired immune-mediated demyelinating neuropathy with slowly progressive asymmetric weakness, fasciculations and cramping. Although significant sensory involvement is usually lacking in MMN, this feature might be present in the Lewis–Sumner variant of the disease. Many patients are at least initially well managed with IVIG infusions. In most neurological diseases with efficacy of IVIG, patients respond equally well to PE. Yet, this is clearly different in MMN, where PE and even more GS might worsen the course of the disease.56–59

Paraproteinemic neuropathy is another rare subtype of inflammatory neuropathy. Here, demyelination is often associated with the presence of clonally restricted IgM antibodies directed against myelin. Theoretically, PE might thus lead to elimination of these pathogenic antibodies and promise a significant clinical effect. In clinical practice however, results from studies are modest at best, whereas some case reports describe a good response to PE.60 These ambiguous results might be explained by the low number of patients in the trials or inclusion of cases with monoclonal gammopathy of unknown significance (MGUS), which might constitute a different disease entity. Also IgM antibodies might be much more difficult to remove from tissue compartments as a result of their high binding affinity.

Finally, PE might not only be of value for patients suffering from immune-mediated neuropathies, but also immune-mediated diseases of the muscle. Yet, for the application of PE in acute exacerbations of polymyositis or dermatomyositis, available data are limited to some case series and retrospective studies.61


Although plasmapheresis has been established as effective in many autoimmune disorders, there are still issues to be addressed over the coming years. Immunological investigations and experimental studies should further clarify the mechanisms of PE action. Although the elimination of antibodies is a conceivable mechanism of action, these factors also influence cellular crosstalk. Thus, the modulation of cellular immune responses after elimination of humoral factors needs to be characterized in more detail. Here, Fc receptor bearing cells, including macrophages and NK cells, are in the centre of interest.

Further trials on the value of immunoadsorption are urgently needed, which might be of special interest as escalation therapy for MS relapses. Whereas in myasthenia gravis, some data support the superiority of immunoadsorption regimes as compared with conventional plasma exchange, the value of immunoadsorption in other autoimmune diseases is less clear. Randomized and double-blind head-to-head studies are warranted to prove equivalence of immunoadsorption to therapeutic plasma exchange. So far, the USA clinical trial registry ( only lists one study testing the value of plasma exchange for the elimination of natalizumab in MS patients. Even here, when comparing individual case reports, immunoadsorption may be superior.62,63

Finally, some retrospective studies already evaluate appropriate long-term follow-up therapies after plasma exchange in MS patients. From these data, it is clear that the need for plasma exchange in MS relapses should coincide with the initiation of a disease modifying therapy. Alternatively, an existing long-term therapy should be critically re-evaluated for the need of therapy escalation. In particular, immunotherapies targeting B cells or plasma cells as the source of pathogenic antibodies are of special interest as follow-up therapies. Although mitoxantrone might lead to B cell apoptosis,64 the B cell depleting anti-CD20 monoclonal antibody showed some very promising effects on MRI markers of disease in two recently published phase II studies65,66

As the ultimate goal, the development of suitable biomarkers for PE responders requires special attention. Here, proteomic profiling of sera and eventually also CSF from MS patients might help to identify suitable markers that need to be tested in clinical practice.