Acute-onset chronic inflammatory demyelinating polyradiculoneuropathy

About 5% of patients initially diagnosed with GBS later turn out to have acute-onset chronic inflammatory demyelinating polyradiculoneuropathy (A-CIDP) and should be treated as for CIDP.²³ Clinical and laboratory features possibly related to A-CIDP are discussed (PICO 6). A few of these patients, especially those not responding well to IVIg or PE, may have antibodies against nodal–paranodal antigens (like NF155, Caspr1 or CNTN1), which define a different disease (autoimmune nodopathy) (PICO 3).

Pathogenetic subtypes

Largely based on a combination of clinical, electrodiagnostic and morphological features, GBS has been classified into different pathogenetic subtypes: acute inflammatory demyelinating polyradiculoneuropathy (AIDP), acute motor axonal neuropathy (AMAN) and acute motor and sensory axonal neuropathy (AMSAN).¹⁴ Multiple papers on this subject in humans and animal models have been published.^24-27

In AMAN cases, the pathophysiology is either (a) reversible conduction failure, due to reversible axonal conduction block at the nodes of Ranvier or the motor nerve terminal without axonal degeneration, followed by rapid recovery,^{28, 29} or (b) extensive axonal degeneration associated with inexcitable nerves and a worse outcome.³⁰ In AIDP cases, there is conduction slowing and/or conduction block, associated with segmental demyelination with or without secondary axonal degeneration, usually followed by recovery through regeneration of myelinating Schwann cells.²⁷

Preceding infections are relevant, especially Campylobacter jejuni, which can induce cross-reactive antibodies that bind to human gangliosides GM1 or GD1a at the nodes of Ranvier or motor nerve terminal, activate complement and disrupt sodium-channel clusters and axoglial junctions, leading to nerve conduction failure.^{3, 31} C. jejuni infections are particularly associated with AMAN and/or motor GBS, and worse outcome³² (see: PICO 14). The demyelinating subtype is predominant in all global regions tested, but AMAN and motor GBS are more frequent in Asia (especially Bangladesh) than Europe and America.^{4, 31}

Some important issues remain only partially resolved. (a) The best electrophysiological criteria to distinguish AIDP from AMAN are currently unknown, and all electrophysiological criteria have limitations. This led to the introduction of the concept of ‘nodo-paranodopathy’ to describe situations in which nodal membrane injury of the axon or through detachment of terminal myelin loops (or both) accounts for acute conduction failure, in the absence of segmental demyelination.²⁹ (b) Anti-GM1 antibodies can also be found in some cases classified as AIDP,³³ although much less commonly, and both subtypes of GBS can be mediated by complement-fixing anti-GM1 antibodies.³⁴ (c) There is some overlap, as both motor GBS and AMAN may have electrophysiology mimicking demyelination, AIDP often also has axonal degeneration, and there appears to be a spectrum of sensory involvement in AMAN and AMSAN.³⁴

The TF acknowledges the relevance of studying the pathophysiological mechanisms leading to different subtypes of GBS. However, the TF deliberately avoided giving or endorsing specific diagnostic criteria for these entities, because there is currently no gold standard to choose between the various published criteria, and the distinction between these subtypes of GBS currently does not affect clinical management.

Pain and fatigue

Pain often occurs in GBS, at any time in the disease course, even before the onset of weakness, and may be severe.³⁵ Fatigue is an important and frequently occurring residual complaint that may persist when weakness has recovered.^{36, 37}

Considerations supporting the GPP (supporting information)

Evidence summary: In patients with suspected GBS, the diagnosis of GBS is more likely if there is a history of recent (within the previous 6 weeks) diarrhoea (sensitivity 13%–18%, specificity 89%–100%),^49-53 Campylobacter infection,⁵⁴ respiratory infection (sensitivity 21%–68%, specificity 59%–98%),^49-53 fever (specificity up to 100%)^{49, 52} or influenza-like illness (specificity up to 100%).^{48, 53} However, around one third of GBS patients report none of these. Case–control studies showed that GBS is associated with infections (C. jejuni, cytomegalovirus (CMV), Epstein–Barr virus, Hepatitis E virus, Zika virus), and with Mycoplasma pneumoniae in children.^{14, 54-57} Although large case–control studies are lacking, from the epidemiological and cohort studies that exist, GBS is not or marginally measurable increased in incidence after SARS-CoV-2 infections.^{48, 58} As this PICO assesses the clinical history prior to the onset of GBS, infections with non-specific symptoms that typically require serological diagnosis fall outside the scope of this Guideline. Specific testing for these is generally not clinically useful, as laboratory results likely come after the diagnosis has been made, but may be warranted in selected cases.⁵⁹ There is some evidence about surgery related to GBS as retrospective and epidemiological studies reported an association between GBS and recent surgery (sensitivity 39%, specificity 98%), with a stronger association for surgery on bones or digestive organs.^60-62

Although not assessed in controlled studies, the risk of GBS may be increased shortly after receiving select biological drugs affecting the immune system,^{52, 63} especially immune checkpoint inhibitors or anti-tumour necrosis factor alpha agents.^{5, 64-66}

Rationale: Asking about these antecedent events may increase diagnostic confidence and is generally more useful when present than absent.

Vaccinations: Prior vaccinations were not originally part of this PICO. However, the TF considered it helpful to comment, although did not systematically review the evidence. Most vaccinations have no proven association with any increased risk of GBS.^{3, 45} The risk of GBS after standard seasonal influenza vaccine (trivalent or quadrivalent) is about one excess GBS case per million vaccines, though historically higher after the 1976 monovalent H1N1 ‘swine’ influenza vaccine.⁴⁶ In a retrospective questionnaire study in 245 patients with previous GBS, subsequent influenza vaccination gave no significant increased risk of recurrence of GBS.⁶⁷ After a recombinant herpes zoster vaccine, there was possibly a similar very small increased risk of GBS.⁴⁷ Large studies of SARS-CoV2 vaccines found a small excess risk of 4–7 excess cases of GBS per million after adenovirus-vector vaccines,^{43, 44, 48} without association with any specific clinical features,⁴⁴ but no increased risk after pegylated mRNA vaccines.⁴⁴ When assessing a patient with GBS, a history of recent vaccination is likely of negligible importance towards making the diagnosis of GBS. For all these diseases, the benefits of vaccination (reducing morbidity and mortality from infection and infection-associated GBS) exceed any extremely small increased risk of developing vaccination-induced GBS.

Considerations supporting the GPP (supporting information)

Evidence summary: CSF examination in patients suspected to have GBS classically shows an increased CSF protein and normal CSF white cell (leucocyte) count.⁴ Diagnostic sensitivity of an increased CSF protein depends on the time CSF is examined after onset of weakness.¹⁵ In a study with over 1000 GBS patients, 50% had an elevated CSF protein at fewer than 3 days from onset (median protein level 0.45 g/L), and 84% at more than 7 days from onset (median protein level 0.98 g/L).⁴ A CSF white cell count <5 cells/μL was found in 80%, a mildly elevated white cell count (5–50/μL) in 19% and more than 50 white cells/μL in 2% of patients.⁴ The specificity of a raised CSF protein is unknown, and it does not rule out some mimics of GBS (Table 2). Normal values of CSF protein are higher in older people.^{68, 69} Both CSF protein and cell count may be artefactually raised after IVIg.⁷⁰

Rationale: Although CSF examination is often performed, because increased protein may support the diagnosis and to exclude mimics with increased cell count, the diagnostic accuracy is unknown.

Considerations supporting the GPPs (supporting information)

Evidence summary: Serum antibodies that have been tested in GBS cases include GM1,^71-75 GM1b,⁷⁶ GD1a,⁷⁷ GQ1b,^78-83 GalNAc-GD1a,^84-86 GT1a,⁸⁷ GD1b, GD3, O-GD3, GT3, O-GT3,⁸⁸ sulfatide,^{89, 90} galactocerebroside,^{91, 92} CNTN1,⁹³ NF155 and NF 186,^{94, 95} cardiolipin,⁹⁶ LM1,^97-99 sulphated glycolipids,⁸⁹ P0 and PMP22.¹⁰⁰ The test accuracy varies depending on GBS subtype, tested antigen and control group. Anti-GM1 IgG antibody sensitivity for the whole group of GBS patients ranges from 20% to 69% and varies between geographical regions.^{71-75, 101} Anti-GM1 specificity is reported to be high (94%–98%) in GBS compared with healthy controls and other neurological diseases.^{25, 54, 73, 102} In GBS, the sensitivity of anti-ganglioside antibody panels is reported to vary between 32% and 64%, depending on the presence of a recent infection or GBS subtype (AIDP or AMAN).^103-105 For MFS, sensitivity for anti-GQ1b antibodies is high (88%–100%),^79-81 with a very high specificity (100%).⁸³ Especially, when there is some clinical doubt and test results can be obtained within reasonable time, testing for anti-GQ1b antibodies is considered helpful. The INCAT group recommended using standardised methods for ELISA to test for anti-ganglioside antibodies.¹⁰⁶ Several studies looked for antibodies against microorganisms (Zika virus,^{107, 108} CMV,^{109, 110} C. jejuni,^{111, 112} Haemophilis influenzae¹¹³) with varying sensitivity and specificity (PICO 1). For patients initially diagnosed with GBS but who are subsequently considered to have an autoimmune nodopathy, see also the EAN/PNS CIDP Guideline for further details.^{114, 115}

Rationale: To diagnose motor-sensory GBS, there is currently no indication to test for anti-ganglioside antibodies because of their low–moderate sensitivity and frequent delay in reporting results of antibody assays beyond the therapeutic window. GQ1b antibody testing is useful when MFS is suspected.

Considerations supporting the GPPs (supplementary material online)

Evidence summary: Prospective and retrospective studies that evaluated up to 84 patients with clinically suspected GBS or up to 66 AIDP cases with a variable number of controls concluded that numerous nerve conduction parameters are abnormal early after disease onset.^116-131 However, some studies included small numbers of patients and/or did not include controls. Sensitivity and specificity vary according to electrodiagnostic criteria and the control groups used. Absence of H-reflexes has a high sensitivity (95%–100%) for AIDP^{105, 121} with 33% specificity.¹¹⁹ It was concluded that the presence of H-reflexes renders the diagnosis of GBS unlikely.^{118, 122} A sural sparing pattern is present early in the disease course with 16%–100% sensitivity^{116, 118, 119, 122, 124} and 91%–98% specificity.^{116, 118} Sural sparing pattern may occur in all subtype classifications (AIDP, AMAN, unclassified) and in MFS.¹²⁵ Distal CMAP duration prolongation can especially be found in AIDP¹¹⁷ and was reported to have 32%–88% sensitivity with high specificity (91%–100%).^{118, 117, 128} Decreased motor nerve conduction velocities and increased distal motor latencies were found in 78% of AIDP cases.¹²² If a patient meets published criteria for AIDP,^{24, 132} this may support the diagnosis of GBS, but the TF considered the evidence of insufficient certainty, and does not advise electrodiagnostic subtyping in clinical practice. Indirect discharges (often multiple and resembling A-waves) are common in AIDP (sensitivity 59%–100%),^{121, 126, 133, 134} with a reported specificity of 73%, as these have also been reported in AMAN.¹¹⁸ Facial nerve direct responses and blink responses are often abnormal early in the disease, but specificity is low.^{121, 123, 124, 135} Nerve root stimulation can detect very proximal conduction block by electrical stimulation of lumbar roots¹³⁶ or magnetic stimulation of the ulnar nerve,¹²⁷ but there is only low certainty evidence for this. In MFS, sensory and motor conduction abnormalities are present in >50% of cases (high sensitivity but low specificity), and sural sparing pattern has low sensitivity but high specificity.^129-131

Rationale: Electrodiagnosis is an important test that is relatively easy to conduct and that may help to diagnose GBS. Since there is no consensus on definitions for electrophysiological subtyping between AIDP, AMAN and AMSAN,^{24, 132, 137-139} and currently it has no impact on management and treatment, the TF decided that electrodiagnostic subtyping was not further considered in this Guideline.

Considerations supporting the GPPs (supporting information)

Evidence summary: Studies identified on nerve MRI in GBS were small (range 11–40 patients) and uncontrolled; in these circumstances showing high sensitivities (82%–95%), but with unknown specificity.^140-144 Studies identified on nerve US in GBS were small (range 17–50 patients) and were mostly controlled but used highly selected patient populations and often compared with healthy controls only. Furthermore, most studies lack objective cut-off values for abnormalities. Within these significant limitations, they report sensitivities ranging from 47% to 95% and specificities ranging from 36% to 91%.^145-153

Rationale: Nerve MRI or US should only be considered if the diagnosis of GBS is uncertain, possibly to rule out other causes. Abnormal nerve MRI and US may help to localise the pathology to the nerve roots, but the tests lack specificity and do not rule out GBS when normal. Further research to define cut-off values and to evaluate the specificity of nerve MRI and US in patients suspected to have GBS and its mimics are indicated.

Considerations supporting the GPPs (supporting information)

Evidence summary: In a prospective study of 170 patients initially diagnosed as GBS, 16 (9%) had a TRF and another 8 (5%) were subsequently diagnosed with A-CIDP.²³ A-CIDP was reported more likely (>5%) if there are marked sensory disturbances on examination (sensory ataxia is the most specific), but facial or bulbar weakness and preceding infections were less frequently observed in A-CIDP compared with GBS.^{149, 154, 155} Confirmation of these results however warrant additional studies in larger numbers of patients. A-CIDP is likely in patients initially diagnosed with GBS, if there is further worsening after 8 weeks from onset (sensitivity 100%, specificity 92%), or when there are three (or more) TRFs (episodes of worsening following treatment-induced improvement/stabilisation) (sensitivity 52%, specificity 96%).²³ Reduced MNCV < 90% of lower limit of normal (or <85% if small distal CMAP) was more frequently found in A-CIDP than in GBS.^{23, 149, 155} Patients with A-CIDP tested with nerve US within 4 weeks of onset had greater enlargement of peripheral nerves compared with those with GBS (sensitivity 88%, specificity 84%).^{147, 149} The absence of IgG anti-ganglioside antibodies has a sensitivity of 96% and specificity of 35% for A-CIDP compared with GBS.^{23, 24, 147, 155} Some patients suspected to have A-CIDP may have an autoimmune nodopathy.¹⁵⁶ These patients have a poorer response to conventional therapies for CIDP, and there is anecdotal evidence that these patients may response to rituximab (see EAN/PNS Guideline CIDP).^{114, 115}

There is no evidence from an RCT, but observational data indicate that a repeated course of IVIg or PE can be effective in case of a TRF.^{157, 158}

Rationale: It is important to diagnose A-CIDP because treatment differs from GBS.^{114, 115} A-CIDP however should not be over-diagnosed in severely weak patients with slow or no improvement. If in doubt, the presence of muscle wasting and denervation on electromyography indicates that GBS-related axonal degeneration is more likely than A-CIDP.

Considerations supporting the GPPs (supporting information)

Evidence summary:

mEGRIS prognostic model: An individual patient's risk of requiring mechanical ventilation can be estimated by using EGRIS^{160, 161} the updated version of EGRIS,¹⁶² and now with the modified (and simplified) version mEGRIS.¹⁵⁹ An advantage of mEGRIS is that it can be used across the full spectrum of GBS, including mild cases and variants, and in patients from different regions. The original EGRIS is based upon 397 GBS patients in a prospective cohort, and 191 patients in a validation cohort, and has good discriminative ability (area under the curve [AUC] 0.84) (moderate certainty evidence). EGRIS was externally validated in a retrospective cohort of 177 Japanese patients.¹⁶³ The updated version of EGRIS is based upon 1023 patients from the International GBS Outcome Study IGOS (Europe/North America n = 842, Asia n = 104, other n = 77), aged ≥6 years, and includes GBS variants and those with mild symptoms.¹⁶² In this cohort, 104 (10%) required mechanical ventilation within the first week from study entry.

Individual factors were assessed for the certainty of evidence to predict respiratory failure (requiring mechanical ventilation) based on both controlled and observational studies. Predictors of respiratory failure (high certainty evidence) are: (1) Shorter time from onset to admission,^{160, 164, 165} (2) Bulbar involvement,^{102, 160, 164, 166} and (3) MRC sum score <20/60.^{161, 167} Probable predictors of respiratory failure (moderate certainty evidence) are: (1) neck muscle weakness,^167-169 (2) greater GBS disability score,^{160, 164, 170} (3) lower vital capacity and^{164, 166, 171} (4) hypoalbuminemia.¹⁷² Factors that may predict respiratory failure (low certainty evidence) are: (1) inability to lift elbows,¹⁶⁵ (2) inability to stand,¹⁶⁵ (3) inability to cough,¹⁶⁵ (4) dysautonomia,¹⁰² (5) lower single breath count (may be more an indicator than a predictor),¹⁶⁶ (6) increased liver enzymes,¹⁶⁵ (7) lower proximal/distal CMAP ratio,¹⁷¹ AIDP versus AMAN or intermediate GBS subtype^{163, 173} and (8) longer phrenic nerve latency.¹⁶⁶ The modified version (mEGRIS) is based upon the first 1500 patients included in the IGOS.¹⁵⁹ Of these, 1133 (76%) patients met the study criteria. Independent predictors of MV were a shorter time from onset of weakness until admission, the presence of bulbar palsy and weakness of neck flexion and hip flexion. mEGRIS was based on these factors and accurately predicts the risk of MV with an AUC of 0.84 (0.80–0.88). The model was internally validated within the full IGOS cohort and within separate regional subgroups, which showed AUC values of 0.83 (0.81–0.88) and 0.85 (0.72–0.98), respectively.

Assessment of important risk factors: The TF found 10 studies assessing predictive factors of the need for mechanical (endotracheal) ventilation. In a large retrospective series of 722 adults, 313 were ventilated and six important predictors (see Figure 2) of subsequent ventilation were identified¹⁶⁵: (1) <7 days from symptoms to admission (OR 2.51); (2) inability to cough (OR 9.09); (3) GBS disability grade ≥4 (wheelchair bound or bedridden) (OR 2.53); (4) inability to abduct shoulder to horizontal (OR 2.99); (5) neck flexion weakness (OR 4.34); and (6) abnormal liver function tests at presentation (OR 2.09). If 4 out of 6 predictors were present, mechanical ventilation was required in 85% of patients.

In a subpopulation, it was found that three individual features were strongly associated with the need for subsequent ventilation. If there was neck flexion weakness (OR 5.00), a duration between onset and nadir of <7 days (OR 5.00) and an FVC of <60% predicted (OR 2.86) were all present, then mechanical ventilation was required in >85% of patients. Ventilation was also associated with the presence of facial, bulbar or clinical markers of autonomic failure.¹⁷⁴ Co-existing CMV (OR 8.81, CI 2.34–33.1) or herpes simplex virus (HSV) infections (OR 4.83, CI 1.16–20.1) were associated,¹⁷⁵ and pulmonary infection (abnormal chest x-ray) is possibly associated (49 vs. 29%, p = .06) with a subsequent need for ventilation.¹⁷⁴

Clinical assessment of respiratory function: Quantitative indicators of the likelihood of a need for ventilation are FVC of <20 mL/kg, MIP < 40 cm H₂O, MEP < 30 cm H₂O or a recorded reduction of >30% in FVC from baseline.¹⁷⁴ A decline of FVC of >50% was associated with mechanical ventilation within 36 h, and a drop in the FVC < 1 L was associated with artificial ventilation within 18 h.¹⁷⁶ SBC < 20 predicted the need for subsequent ventilation.¹⁶⁶

Risk of prolonged mechanical ventilation: Risk factors for prolonged mechanical ventilation (14 days or more) in individuals with GBS have been studied in a cohort of 552 participants (recruited from RCTs and observational studies).¹⁷⁷ Prolonged mechanical ventilation in GBS may be predicted by (1) inability to lift arms¹⁷⁷ or (2) axonal GBS or inexcitable nerves (low certainty evidence).

Complications after ICU discharge: It was reported that two thirds of deaths of GBS patients occur in the period following ICU discharge and during the recovery phase, most frequently from respiratory or cardiovascular complications.¹⁷⁸

Rationale: It is very important to regularly assess indicators of progressive respiratory impairment (ventilatory insufficiency due to muscle weakness and also infections) and autonomic dysfunction in the early and progressive phases of GBS to avoid emergency intubations or otherwise life-threatening situations. The TF has constructed a flow chart that may aid (Figure 2).

Considerations supporting the recommendations (supporting information)

Evidence summary: The TF considered the effectiveness of PE compared with supportive treatment only (evidence from 6 RCTs including 649 patients), its rather limited adverse effects and the high disease burden, crucial in decision making.^{179, 183-188} PE reduced the time to regain ability to walk with aid, reduced the time to onset of motor recovery and increased the proportion of patients who recovered the ability to walk unaided (low to moderate certainty evidence). After 1 year, full recovery of muscle strength was somewhat more likely and severe residual weakness somewhat less likely in patients treated with PE (moderate certainty evidence). Moderate certainty evidence showed that the effect of PE was only demonstrable when started within the first 4 weeks of onset of weakness, and most effective when started within the first week of onset of weakness.^{185, 186}

The recommended total number and exchanged volume of PE sessions vary between studies.¹⁷⁹ The advice of the TF mainly originates from three large studies.^185-187 Two trials compared different numbers of PEs related to the level of GBS disease severity.^{186, 187} The largest study in 161 ventilated GBS patients showed that six PE sessions (each of 40 mL/kg or 1.5 plasma volumes) may be as effective as four PE sessions (low certainty evidence).¹⁸⁷ In 304 moderately severe GBS patients (not ventilated), four PE sessions (in total 4 × 3.75 L plasma removed) may be superior to two PE sessions (low to moderate certainty evidence).¹⁸⁷ In the 91 GBS patients still able to walk or stand up unaided, two PE sessions appeared to be better than no PE,¹⁸⁷ but this is considered to be a low certainty evidence. In the North American PE trial (including 245 GBS patients), in total, 200–250 mL plasma/kg was removed in five exchanges over 7–14 days.¹⁸⁵ According to the most frequently used PE schedules in the largest trials, the TF recommends four to five exchanges of about 3 L plasma each, with a total removal of 12–15 L plasma (dependent on body weight and possible side effects).^{158, 179} There is no RCT on the effect of PE (or IVIg) in very mildly affected patients (GBS-DS 1) within the first 2 weeks from onset of weakness, nor in (very) mildly affected patients (GBS-DS grade 1–2) 2–4 weeks after onset of weakness. In one trial, 57 participants were randomly allocated to receive PE with albumin and crystalloids, and 52 participants received fresh frozen plasma as the replacement fluid. No significant differences were found between the two arms in any of the outcomes. Although fibrinogen and prothrombin decreases were greater in participants receiving albumin, adverse events were more frequently found when fresh frozen plasma was used as replacement fluid.^{186, 189, 190}

A single report of a small number of patients suggested that small volume plasma exchange (SVPE)—a technique in which multiple small blood volumes are sequentially exchanged—can be used relatively safely without requiring specialised equipment (very low certainty evidence).¹⁹¹ Efficacy data on SVPE are not available.

Rationale and implementation considerations: RCTs showed that PE is an effective treatment for patients with GBS. PE is a less favourable option than IVIg in young children due to limited vascular access and the child's possible fear of the procedure. Patients with severe cardiovascular autonomic instability have a relative contra-indication to PE with discontinuous filtration machines because of the large volume shifts and possible blood pressure changes, but these are a lesser problem with continuous flow machines.

In very mildly affected patients (GBS-DS grade 1) within the first 2 weeks from onset of weakness, or in mildly affected patients (GBS-DS grade 1–2) without additional difficulties (like swallowing or autonomic disturbances) within 2–4 weeks from onset of weakness, there is no indication that PE (or IVIg) is beneficial. Because treatment may induce side effects and is expensive, it is not advised (GPP) to start treatment in these patients. PE requires local availability of specialised equipment and skills, not readily available in all hospitals. In under-resourced parts of the world where standard care for GBS (including ICU facilities, PE or IVIg treatment) is unaffordable or unavailable for many patients, efforts to find innovative treatment alternatives should be encouraged. Whether SVPE is as effective as standard PE and partially overcomes this problem requires additional studies.

Considerations supporting the recommendations (supplementary material online)

Rationale: IVIg is generally associated with few adverse events and is readily available in most hospitals. PE requires special facilities, good intravenous access and has a slightly higher adverse event (AE) rate (at least when non-continuous flow apheresis was used). In children, the relative burden of PE may be a reason to prefer IVIg. Due to the high costs, patients in some countries may be unable to afford either treatment.

Considerations supporting the recommendations (supporting information)

Evidence summary: This recommendation is based upon one RCT comparing IVIg or PE, with PE immediately followed (irrespective of the prognosis) by IVIg in 249 participants.¹⁹³ Treatment with IVIg or PE (standard treatment schedules) was started within 14 days of GBS onset. In the combined treatment group, the patients first received PE, followed by IVIg (0.4 g/kg/day for 5 days), which was started on the day after the last PE (irrespective on the effect PE). No significant differences were found in the three arms of this RCT (moderate certainty evidence). There is no indication that a subgroup of patients would benefit more from the combined treatment with PE followed by IVIg.

A small open-label study in nine selected paediatric patients with severe GBS describes the effect of alternating PE/IVIg (one PE session, immediately followed by IVIg (0.4 g/kg), both repeated alternatively five times, ‘Zipper method’).²⁰¹ This unconventional treatment schedule washes out IVIg very rapidly by repeating the PE sessions. To assess whether it may have an objective positive clinical effect requires an appropriate-scale RCT.

There are no appropriate data on the possible effect of giving the alternative treatment (PE or IVIg) in patients who further deteriorate despite standard treatment with IVIg or PE.

Rationale: As the combined treatment of PE followed immediately by IVIg (in an RCT irrespective of the clinical course after PE) has probably no increased effectiveness, potential additional (limited) adverse events and an increase in costs, this treatment schedule is not recommended.^{180, 193} PE and IVIg seem equally effective, and there are currently no subgroups identified which do better after either PE, IVIg or PE followed by IVIg. Starting PE immediately after IVIg washes out the treatment that has just been infused, which theoretically seems unattractive and expensive. It is not excluded that some patients who do not do well after their first treatment might possibly do better after a different treatment. However, based on a lack of trial evidence, and the arguments mentioned above, the TF does not advise giving the alternative treatment as a standard procedure in patients who further deteriorate despite a course of IVIg or PE.

Considerations supporting the recommendation (supporting information)

Evidence summary: One small open pilot study, a prematurely stopped RCT and a very small retrospective cohort study were identified.^{195, 202, 203} The open randomised pilot study included 45 GBS patients, of which 11 received PE, 13 IA and 21 IA followed by IVIg, which may suggest some advantages for the sequential regimen.²⁰² The RCT aimed to recruit 279 patients but was prematurely stopped after including 23 patients treated with IVIg, 26 with PE and 18 with IA.¹⁹⁵ No significant benefit was found for IA alone, nor for the combination of IA followed by IVIg (very low certainty evidence).^{180, 195, 202}

Rationale: Because IVIg is a proven effective treatment for GBS and there is insufficient data on the effectiveness of IA, the TF concluded that there currently is insufficient information to support the use of IA for treatment of GBS. IA is expensive and in most countries has only limited availability. Practical experience suggests that IA may be safe and effective; however, this requires proper validation.

Considerations supporting the recommendations (supporting information)

Evidence summary: The TF considered the probable lack of efficacy of IVMP (moderate certainty evidence from a trial with 242 patients),^{181, 204} the probable lack of efficacy of the combination of IVIg and IVMP compared with IVIg and placebo (moderate certainty evidence from a trial with 225 patients),²⁰⁵ the probable harm (delayed recovery) of oral corticosteroids (low certainty evidence from 4 trials with a total of 120 patients) using various oral regimens of the equivalent of 40 mg prednisolone daily for at least 2 weeks,^{38, 181, 205-208} and the high certainty evidence of adverse effects (diabetes more common), despite that hypertension was less common in the corticosteroid-treated patients, which is crucial in decision making.^{204, 205}

Rationale: Corticosteroids are inexpensive, but there is no trial evidence to support the use of corticosteroids in patients with GBS. Intermediate high dosages of oral corticosteroids may even harm GBS patients.

Considerations supporting the recommendation (supporting information)

Evidence summary: In two small phase 2 trials (including a total of 41 patients), no beneficial effects of eculizumab could be demonstrated (very low to low certainty evidence).^{182, 213, 214} Both trials used the same protocol. One trial in seven patients was prematurely ended because recruitment was very slow.²¹⁴ The Japanese eculizumab trial (JET-GBS) in 33 patients (randomised 2:1 for eculizumab compared to placebo) did not show significant effects on pre-specified outcome measures; however, more patients were able to run after 6 months compared with placebo.²¹³

Rationale: The lack of demonstrated efficacy, the known adverse effects (all patients had some adverse effects) and the high cost currently result in a weak recommendation against eculizumab. Further trials are awaited.

Recommendations

Considerations supporting the recommendation (supporting information)

Evidence summary: Only small, very low certainty RCTs and case series evaluating these drugs (see: recommendation 1) were identified, for which no clinically important differences in any of the outcome measures could be demonstrated for any of the interventions described.^{182, 215-224} Minor clinical benefit or harm cannot be excluded. As these treatments potentially can have side effects and important clinical benefit is unlikely, the TF recommended strongly against using these treatments.

The TF considered that the effectiveness of neuromuscular electrical stimulation during early rehabilitation phase of disease is uncertain, based on one small RCT (12 participants).^{225, 226} Both the effectiveness and safety of 3,4-diaminopyridine (seizures have been observed in other studies) are considered uncertain (one observational study, four GBS patients).^{227, 228} In a cross-over study (seven participants) with chronic disability after GBS (over 12 months after onset), some improvement of motor function after 3,4-diaminopyridine was found.²²⁹ Because of the small trial size, it is not possible to draw a conclusion about the effectiveness in patients with long-term stable weakness after GBS.

Rationale: Given the costs and safety profiles, and the lack of proven clinical benefit, the TF currently recommended against alemtuzumab, brain-derived neurotrophic factor, CSF filtration, cyclophosphamide, interferon beta 1a, muronomab-CD3, mycophenolate mofetil and tripterygium polyglycoside. Because of the uncertainty about an effect on motor function and the potential side effects, the TF weakly recommended against using 3,4-diaminopyridine for the treatment of GBS. Further research needs to be conducted before a recommendation about the possible effectiveness of neuromuscular electrical stimulation can be given.

Considerations supporting the recommendation (supporting information)

Evidence summary: No RCTs on physiotherapy, occupational therapy or rehabilitation treatment in the acute phase of GBS were identified. Only a small randomised study (79 patients) in the chronic phase of GBS was found. This study indicated that a high-intensity intervention may reduce disability more than a low-intensity intervention, but this requires additional studies.²³⁰ Despite the absence of RCTs, TF is unanimous in their support of the use of physiotherapy, occupational therapy, speech or respiratory therapy (when indicated) and rehabilitation treatment in the various phases of GBS.

Considerations supporting the recommendations (supporting information)

Evidence summary: There are only small short-term studies in the acute phase of GBS available that provide low certainty evidence. Carbamazepine (100–200 mg three times daily) was beneficial in two small trials versus placebo (low certainty evidence).^{231, 232} Two small RCTs support the use of gabapentin (5 mg/kg or 300 mg, three times a day) for the treatment of neuropathic pain (low certainty evidence).^{232, 233} It has been argued that pain in the acute phase in some patients may be due to proximal localised endoneurial inflammatory oedema that may be responsive to corticosteroids.²³⁴ Methylprednisolone (500 mg for 5 days) probably offers no clear benefit to the treatment of either pain or disability (moderate certainty evidence).^{35, 235} Guidelines for treatment of chronic neuropathic pain are available.^{236, 237}

Rationale: There currently is no clear indication that neuropathic pain in GBS should be treated differently from neuropathic pain in other peripheral nerve diseases.

Considerations supporting the recommendations (supporting information)

Evidence summary: Only one small trial (22 patients) was identified that compared pranayama yoga (breathing practices) and meditation as add-on therapy to usual care.²³⁸ The TF is uncertain whether pranayama yoga and meditation as add-on therapy influences sleep quality and pain (very low certainty evidence), or changes in daily activities (low certainty evidence). There is no evidence that patients with GBS may benefit from activities that improve emotional well-being more or less than the general population.

Rationale: Although there is no evidence to show that patients with GBS may benefit from yoga (in all its forms) and other complementary therapies, patients who have had GBS may find that careful introduction of one of these activities may be worthwhile to try.

Considerations supporting the recommendation (supporting information)

Evidence summary: There is low certainty evidence from a cross-over trial in 80 GBS patients that there is no effect of amantadine to reduce fatigue, improve quality of life or increase participation when used at standard doses for 6 weeks.³⁶ No other studies appropriately assessing other drugs to treat fatigue in GBS have been found.²³⁹

Rationale: Although amantadine might reduce fatigue in other disorders, trial evidence in GBS did not show a positive effect. Other long-term conditions with associated fatigue are sometimes treated with anti-fatigue medication, including stimulants (modafinil, methylphenidate), androgenic steroids (DHEA) and antidepressants (serotonin reuptake inhibitors), but this has not systematically been studied in GBS.²⁴⁰

Considerations supporting the GPP (supporting information)

Evidence summary: An open 12-week study of bicycle exercise training in 16 GBS patients and 4 CIDP patients with severe fatigue, who were otherwise relatively well recovered from their disease, indicated that training was well tolerated and fatigue scores improved.²⁴¹

Rationale: It is very unlikely that exercise would be harmful in physically relatively well-recovered GBS patients. Exercise is often feasible and anecdotal reports from patients, clinicians and unpublished studies suggest that it has at least subjective and maintained benefits.

Considerations supporting the recommendations (supporting information)

Evidence summary:

Risk of poor outcome (inability to walk unaided after 3–6 months): Both controlled and observational studies have been found and were used to identify factors related to poor outcome in GBS.

Predictors for poor outcome (high certainty evidence) are (1) older age,^{164, 168, 171, 242-248} (2) preceding diarrhoea/gastroenteritis,^{102, 167, 245-249} (3) higher GBS disability scores at admission,^{102, 164, 167, 247, 249} (4) lower MRC sum scores at admission^{247, 248, 250} and (5) decreased CMAP amplitude.^{243, 244, 251, 252}

Probable predictors of poor outcome (moderate certainty evidence) are (1) preceding C. Jejuni infection,^{102, 247} (2) severe arm weakness,²⁴⁶ (3) higher EGOS scores,^{161, 249, 253} (4) higher mEGOS scores,^{161, 247} (5) higher GBS disability scores at nadir,^{164, 242, 243} (6) absence of CMAPs¹⁶⁷/Inexcitable motor nerves,^{242, 246} (7) decreased proximal/distal CMAP amplitude¹⁷¹ and (8) hypoalbuminemia.¹⁷² Preceding CMV infection probably does not independently predict poor outcome in GBS at 6 months, but may predict poor outcome at 8 weeks.²⁴⁸ The presence of any preceding infection may not independently predict poor outcome in GBS.^{243, 254}

Factors that may predict poor outcome (low certainty evidence) are: (1) short time from onset to admission,^{242, 244, 247, 248} (2) short duration of active disease,²⁴⁵ (3) upper or lower limb power < MRC grade 3,²⁴² (4) respiratory failure /history of mechanical ventilation,^{242, 244, 250, 254, 255, 164} (5) short time to nadir disability,^{245, 243} (6) lower MRC sum scores at nadir,^{254, 256} (7) being bedbound/having respiratory insufficiency at nadir,²⁴⁵ (8) increased CSF protein,²⁵⁷ (9) denervation with needle EMG,²⁵⁸ (10) recruitment pattern of hypothenar muscles with needle EMG,²⁵⁸ (11) anti-GM1 or GD1a antibodies,¹⁰² (12) absence of IgM anti GM1 antibodies,²⁴⁶ (13) diabetes,²⁵⁵ and (14) BCII glucocorticoid receptor gene polymorphisms.²⁵⁹ Neck muscle weakness may not independently predict poor outcome in GBS.^{250, 254} Increased serum neurofilament light chain (NFL) levels are associated with axonal damage and poor outcome after GBS, but it is not yet known if this is independent of other prognostic factors.^{260, 261}

Prognostic model: Risks of poor outcome for individual patients can be calculated by the mEGOS²⁴⁷ (moderate certainty evidence). An IGOS validation study showed that mEGOS also accurately predicts the inability to walk after 4–26 weeks in patients in countries outside The Netherlands. The region-specific version of mEGOS for patients from Europe/North America enables an even more accurate prediction of the inability to walk unaided for patients in those continents.³²

Rationale: The prognosis of GBS varies greatly. Prognostic models help to estimate the prognosis in an early stage of disease.