Article Text

Download PDFPDF

Pragmatic guide to peripheral nerve disease and the role of clinical biomarkers
  1. Ryan Yann Shern Keh1,2,
  2. Sachit Shah3,
  3. James B Lilleker2,4,
  4. Tim Lavin2,
  5. Jasper Morrow1,3,
  6. Aisling S Carr1,5,
  7. Michael P Lunn1,5
  1. 1 Centre for Neuromuscular Diseases, National Hospital of Neurology and Neurosurgery, Queen Square, University College London Hospitals NHS Foundation Trust, London, UK
  2. 2 Manchester Centre for Clinical Neurosciences, Salford Royal Hospital, Northern Care Alliance NHS Foundation Trust, Manchester, UK
  3. 3 Lysholm Department of Neuroradiology, National Hospital of Neurology and Neurosurgery, Queen Square, University College London Hospitals NHS Foundation Trust, London, UK
  4. 4 Division of Musculoskeletal and Dermatological, Manchester Academic Health Sciences Centre, University of Manchester, Manchester, UK
  5. 5 Institute of Neurology, University College London, London, UK
  1. Correspondence to Dr Michael P Lunn; michaellunn{at}


In clinical neurology practice, there are few sensitive, specific and responsive serological biomarkers reflecting pathological processes affecting the peripheral nervous system. Instead, we rely on surrogate multimodality biomarkers for diagnosis and management. Correct use and interpretation of the available tests is essential to ensure that appropriate treatments are used and adjusted in a timely fashion. The incorrect application or interpretation of biomarkers can result in misdiagnosis and delays in appropriate treatment. Here, we discuss the uses and limitations of such biomarkers and discuss possible future developments.


Data availability statement

Data sharing not applicable as no datasets generated and/or analysed for this study.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


In daily clinical practice, neurologists apply Bayesian thinking to judge and interpret diagnostic information, consciously and subconsciously.1 Each new clinical finding or test result is weighed against the prior likelihood of a given diagnosis. It is essential to have a working knowledge of individual diagnostic test accuracy and how this accuracy depends on the clinical scenario in which it is being used.

A biomarker is defined as ‘a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention’.2 In clinical practice, we use biomarkers for diagnosis, prognosis, or monitoring disease activity and response to treatment. However, we must use biomarkers selectively to obtain the information we need.

Many biomarkers are described with sensitivity and specificity derived from cohorts of deeply phenotyped patients, and often compared with very specific (but sometimes not completely appropriate) control cohorts. When subsequently used for clinical purposes as ‘diagnostic tests’, the test accuracy (the ability to distinguish between a particular disease and health) is often very different. Biomarkers are often used to add confirmation to prior unsubstantiated belief and are sometimes used as the sole justification for a diagnosis, resulting in diagnostic error and potential harm. Alternatively, the recognition of both the utility and the limitations of each biomarker and their judicious use makes them a useful adjunct to the skill of the clinical neurologist.

This review provides a practical guide to the understanding the uses—and perhaps more importantly the limitations—of common biomarkers in diagnosis and management of the patient with neuropathy and discusses potential future developments. The range of investigations used in peripheral nerve disease is extremely broad, and as such, the focus of this review is on the treatable neuropathies and the biomarkers that help within that context. We therefore will not discuss the following areas in detail:

  • Neurophysiological changes in each individual neuropathy.

  • Role of nerve biopsy in investigating neuropathy (see Nathani et al 3).

  • Investigation of hereditary neuropathy (see Rossor et al 4).

  • Investigation of autonomic neuropathy (see Brown5).

  • Investigation of small fibre neuropathy (see Themistocleous et al 6).

General approach in diagnosis of neuropathy

Peripheral nerve disease, like other areas in neurology, is clinically eloquent. The phenotype obtained from a good history and examination often correctly establishes the diagnosis. The tempo of onset, pattern of nerve involvement and progression, presence or absence of motor, sensory, autonomic or cranial nerve deficits, and coexistence of systemic or central nervous system (CNS) involvement are key pieces of information. Electrophysiological testing using mainly nerve conduction studies (NCS) and electromyography (EMG) is an extension of the clinical assessment and should preferably be done at the same clinical appointment, with the neurologist and the neurophysiologist working closely. At this stage, biomarkers (such as blood tests, imaging or nerve biopsy) should be selected to confirm suspected diagnosis, exclude mimics and guide further treatment.

In the common case of late-onset, length-dependent, symmetrical sensory-predominant axonal neuropathy, the main value of NCS/EMG and blood testing is to detect atypical electrophysiological characteristics (significant asymmetry, non-length dependence and demyelinating features) or the presence of a paraprotein, which should prompt the neurologist to arrange for further appropriate clinical investigations.

Electrophysiological testing as a biomarker

The NCS and EMG are not a substitute for careful clinical assessment. Neurophysiological tests are most useful as diagnostic tools and have limited sensitivity in monitoring disease activity or progression. Here we discuss some practically helpful features of neurophysiological testing in neuropathy.

In a new presentation of generalised neuropathy, the NCS/EMG study can determine whether

  1. The primary pathology is likely to be axonal (where nerve action potential amplitudes are reduced with normal or relatively normal conduction velocity).

  2. The primary pathology is likely to be demyelinating (conduction block at non-compressive sites, temporal dispersion or conduction slowing with relatively preserved action potential amplitudes).

    • Note: conduction slowing can occur with disproportionate loss of large, fast-conducting axons and should not be presumed always to represent demyelination.

    • Note: conduction block may be produced by nodal or paranodal pathology rather than segmental demyelination.

  3. The distribution is homogenous or patchy, symmetrical or asymmetrical, and length-dependent or non-length-dependent.

  4. The dysfunction localises to nerve roots, plexus, nerve, neuromuscular junction or muscle in accordance with or contradicting the clinical assessment.7

It is important to remember that even in experienced, skilled hands, there can be interexaminer variability. Several studies of sensory nerve action potential (SNAP) recordings suggest they have poor reliability, with an intraclass correlation coefficient of <0.5, with an intraclass correlation coefficient of >0.8 considered very good, and >0.6 good reliability or concordance.8 9 Even serial recordings from the same operator vary widely with temperature and machinery, by up to 6.7% in conduction velocity and 32.1% in SNAP amplitude in some studies.10 Further diagnostic tests are therefore often needed.

In acute flaccid paraparesis, such as Guillain-Barré syndrome (GBS), characteristic neurophysiological changes of acute demyelinating polyradiculoneuropathy are often not immediately visible. About half of GBS patients do not meet definitive electrodiagnostic criteria (in the form of clear evidence of demyelinating change in two or more nerves) within the first week of symptom onset.11 However, directed testing can be informative if the correct questions are asked. In very early GBS the H-reflex is lost in up to 97% of patients, with absent or delayed F-waves in 84% reflecting early proximal inflammation at the nerve root.11 Also, relative sparing of sural SNAPs with attenuation or loss of upper limb SNAPs has good sensitivity and specificity for GBS, and to a lesser extent for inflammatory neuropathies in general. Serial NCS/EMG testing can be diagnostically useful where these subtle features are unclear12 and for judging prognosis (see further). International guidelines recommend serial testing in these cases.13

From a prognostic perspective, GBS subclassification into conduction slowing (acute inflammatory demyelinating polyneuropathy) and axonal (acute motor axonal neuropathy (AMAN) and acute sensory motor axonal neuropathy (AMSAN)) variants can be helpful. A worse outcome is expected for AMSAN or AMAN compared with acute inflammatory demyelinating polyneuropathy,14 and early acute and severe denervation may imply poor recovery, although some patients experience excellent recovery after a long delay. The neurophysiological GBS type, in conjunction with the Modified Erasmus GBS Outcome Score15 (a simple severity score based on age, preceding diarrhoeal illness and Medical Research Council Sum Score (MRC-SS); see figure 1), is a powerful pair of prognostic tools.

Figure 1

The Modified Erasmus GBS Outcome, reproduced from Walgaard et al.15 Predicted probability of patients being able to walk independently at 4 weeks (black lines), 3 months (red lines) and 6 months (green lines), on the basis of the mEGOS at hospital admission (A) and at day 7 of hospital admission (B), Calculated based on the prognostic factors within the attached table. Grey shaded areas represent 90% CI. mEGOS, Modified Erasmus GBS Outcome; MRC, Medical Research Council.

In patients with initial NCS suggesting axonal change who experience faster than expected improvement, repeat improved NCS can indicate reversible conduction failure, potentially indicating antiganglioside or paranodal antibody binding to nodal/paranodal structures. On single studies, this can be misinterpreted as axonal damage.12 The significance of this finding is mainly prognostic, as recovery is generally much quicker and more complete than ‘typical’ AMAN/AMSAN.

In gradually progressive or chronic neuropathies, NCS abnormalities reflect the changes in large fibre involvement. In patients with evidence of peripheral sensory deficit on examination or significant deafferentation and normal sensory NCS, a preganglionic lesion is most likely. In the rare disorder chronic inflammatory sensory polyradiculoneuropathy, normal peripheral NCS and abnormal somatosensory-evoked potentials localise sensory pathway dysfunction proximal to the dorsal root ganglion. Supportive investigations such as raised cerebrospinal fluid (CSF) protein and evidence of nerve root thickening and contrast enhancement on MRI of the lower spine can help with confirming the diagnosis and considering appropriate immunomodulatory treatment.16 Pure small fibre involvement from drugs, toxins and alcohol, as well as being rare, is not detected with routine NCS. Quantitative sensory testing can aid in the diagnosis of small fibre neuropathy, although the results of these tests depend on patient reporting (and are somewhat subjective) and operator training. Protocols vary between departments, and routine quantitative sensory testing with thermal thresholds cannot distinguish peripheral and central causes of a deficit, which can occur at any point of the small fibre sensory pathway.6 Some departments may also have access to microneurography (where microelectrodes are placed directly within nerve fascicles), a highly promising research tool with potential for future use in small fibre neuropathy diagnosis and clinical practice.6

Using EMG to evaluate generalised neuropathy is complementary to almost all NCS and should almost always be undertaken where there are no absolute contraindications. EMG of multiple appropriately targeted muscles can confirm a length-dependent distal-to-proximal gradient of abnormalities in neuropathy, illustrate excessive denervation suggesting long-standing, gradually progressive genetic neuropathies, and help delineate coexistent radiculopathies. Based on motor unit configuration, the chronicity of changes can also be estimated. Rarely, EMG may identify concomitant unexpected myopathy such as in mitochondrial disease, although in some myopathic disorders (eg, sarcoid and inclusion body myositis) the EMG can appear ‘neurogenic’. Importantly, myopathic changes on EMG have less diagnostic accuracy than neurogenic findings: a large Greek case series found that EMG had 100% sensitivity and 92.9% specificity for neurogenic disorders but only 76.4% sensitivity and 58.8% specificity for myopathies.17

Blood and CSF biomarkers in neuropathy

Targeted biomarker testing in biofluids is helpful in neuropathies when appropriately directed by clinical phenotype. It is useful to consider this from a statistical and epidemiological perspective. The specific peripheral diagnoses for which we have ‘diagnostic’ blood tests (eg, anti-myelin associated glycoprotein (MAG) neuropathies, polyneuropathy, organomegaly, endocrinopathy, monoclonal protein and skin changes (POEMS) and autoimmune nodopathies) are rare, even within a peripheral nerve patient cohort. When overall case prevalence is low, even tests with relatively high specificity, applied to an unselected fashion, can produce significant numbers of false positives, creating difficulties with clinical interpretation.

The presence of a serum autoantibody does not ‘make’ a diagnosis; it only adds weight to the diagnostic jigsaw puzzle. An autoantibody result should lead to a diagnosis in only very few cases, and then only with a significant re-evaluation of the pre-existing features.

Antiglycolipid autoantibodies

Antiganglioside autoantibodies are commonly requested for patients with inflammatory neuropathy. Gangliosides are N-acetylneuraminic (sialic acid)-containing glycosphingolipids, which coalesce with proteins and glycoproteins in cell membranes as part of functional lipid rafts. These rafts contain channels and receptors to facilitate membrane functions such as signal transduction.18 19 The oligosaccharide epitopes of peripheral nerve gangliosides are targeted by serum autoantibodies in GBS, generated through tolerance breakage and molecular mimicry.20 21 Antiganglioside autoantibodies may occur in a range of acute and chronic inflammatory neuropathies.19 IgG antibodies usually occur in acute neuropathies and IgM antibodies occur in chronic neuropathies.19 Table 1 summarises the common associations between antiganglioside autoantibodies and inflammatory neuropathy.

Table 1

Ganglioside autoantibody associations with inflammatory neuropathies

Antiganglioside autoantibodies, apart from the specific case of IgG anti-GQ1b antibodies as discussed further, are of limited practical use in inflammatory neuropathies. The occurrence of detectable antiganglioside autoantibodies in these inflammatory neuropathies varies from 36% to 81%.18 22–24 Specificity is low, with seropositivity in healthy controls as high as 9%25 and false-positive antiganglioside autoantibodies found if patient sera are tested after giving intravenous immunoglobulin. In addition, assays are lengthy and thus results are often delayed beyond diagnosis and therapy: a positive result affirms a clinical diagnosis, while a negative result is often ignored.

In contrast, Miller Fisher syndrome (the triad of ataxia, ophthalmoplegia and areflexia) has IgG anti-GQ1b antibodies in 80%–90% of cases, and IgG anti-GQ1b is highly specific for any inflammatory neuropathy with ophthalmoplegia.26 It is unclear if the phenotypical features result from complex ganglioside distribution or immunological visibility.27

In chronic inflammatory neuropathies, there is an argument for a diagnostic role of IgM anti-GM1 antibodies in multifocal motor neuropathy with conduction block (MMNCB). Anti-GM1 autoantibodies occur in about half of such cases, possibly indicating more severe disease.28 Meta-analysis suggests GM1 autoantibodies improve the probability for having MMNCB (in patients with a chronic, progressive, asymmetrical, distal-predominant pure-motor syndrome with reflex loss and muscle wasting) from 20%–60% to 50%–85%,29 but they also occur in normal people and those with motor neurone disease (over 50% of motor neurone disease patients in some cohorts)30 and other disorders. In an unselected population, Bayesian probabilities predict that more normal people and those with motor neurone disease would have anti-GM1 antibodies than patients with MMNCB because the latter is a rare disease. Autoantibodies against complexes of GM1 and galactocerebroside may be more sensitive (60%–65% vs 40%) with retained specificity (>91%) compared with anti-GM1 autoantibodies alone in MMNCB diagnosis.31

In a chronic ataxic neuropathy with an ophthalmoplegia and an IgM paraprotein, the presence of IgM autoantibodies to gangliosides with two sialic acids in the epitope (disialosyl antibodies, mainly GD3, GD1b, GT1b and GQ1b) is intrinsic to a diagnosis of chronic ataxic neuropathy, ophthalmoplegia, monoclonal protein, cold agglutinins and disialosyl antibodies (CANOMAD). The disease is rare, and the phenotype overlaps with chronic inflammatory demyelinating polyradiculoneuropathy (CIDP).32 33 CANOMAD is worth considering as the presence of the IgM paraprotein may allow justification for rituximab treatment, and small studies suggest these patients may respond.32 33

Autoantibodies to proteins and glycoproteins

Autoantibodies to epitopes around the node of Ranvier (nodal/paranodal antibodies) found in patients with illness resembling GBS or CIDP are an example of diagnostic biomarkers with prognostic and therapeutic impact. These ‘autoimmune nodopathies’ represent a small but significant 3.8%–18.0% of inflammatory demyelinating neuropathies.34–37 Nodal and paranodal antibodies target the neurofascins 155, 140 and 186 (or all together), contactin-1 (CNTN1), and Caspr. There are now reasonably distinct clinical phenotypes described with each of these.34–36

The occurrence of these autoantibodies associated with the right clinical presentation is highly likely to be both relevant and directive of treatment. Box 1 illustrates two cases of typical nodal/paranodal antibody-associated neuropathy.

Box 1

Neuropathies associated with nodal/paranodal antibodies


Subacute, severe motor-predominant neuropathy with reversible conduction failure, sometimes ataxia and tremor, poor response to IVIg.

Case 1

A patient presented with a 3-month history of progressive sensory ataxia and leg weakness. Cerebrospinal fluid protein was elevated and NCS/EMG confirmed a demyelinating neuropathy. NF-155 was positive. The patient responded poorly to IVIg but well to oral corticosteroids.

Learning point 1

Response to IVIg is often poor in NF-155. Response to corticosteroids is variable.

Case 2

A 71-year-old man had been regularly treated with IVIg for an inflammatory neuropathy for many years. He also had a tremor refractory to medical treatment and nephrotic syndrome due to membranous glomerulonephritis. CNTN1 antibodies were positive. He was treated with rituximab.

Learning point 2

CNTN1 antibodies are associated with neuropathy, tremor and nephrotic syndrome.

  • CNTN1, contactin-1; IVIg, intravenous immunoglobulin; NF, neurofascin.

NF155 disease (usually IgG4 subclass antibodies) is associated with an aggressive, distal, motor-predominant neuropathy, usually in younger patients with ataxia, tremor, high CSF protein, and sometimes concomitant CNS demyelination.34–36 Patients with anti-CNTN1 antibody-associated neuropathy are older, with rapid-onset neuropathy with severe weakness, sensory ataxia and early axonal loss.34–36 The anti-Caspr autoantibody phenotype appears to be associated with significant neuropathic pain.34–36 The primary lesion in autoimmune nodopathies is probably reversible conduction failure,35 with poor response to IVIg and generally better than anti-CD20 B-cell depleting therapies like rituximab.34–36 NF140/186 (again, usually IgG4 autoantibodies) occur in older patients with more subacute onset, sensory ataxia, electrophysiological evidence of conduction block and, in some cases, cranial nerve involvement, and concomitant NF155 autoantibody positivity can be seen.38 Response to IVIg in these is generally good.38

A few patients have pan-neurofascin (positive NF155 and NF140/186) IgG1 antibodies with rapidly progressive tetraplegia and cranial nerve deficits, more common autonomic dysfunction and respiratory involvement, and frequent nephrotic syndrome.39 They respond poorly to IVIg, plasma exchange and corticosteroids, but anti-CD20 therapy is associated with improvement.39 Rituximab is now commissioned for these diseases in the UK.

Paraproteins and associated antibodies

The association between paraprotein and neuropathies is well known, but by no means are all paraproteins relevant to the neuropathy with which they co-occur. Most identified paraproteins are a monoclonal gammopathy of undetermined significance (MGUS). MGUS is common and prevalence increases with age (1% at 50, 3% at 70 and up to 10% over 80 years). Non- IgM MGUS tends not to change management of otherwise straightforward CIDP33 40 apart from prompting consideration of haematology referral for paraprotein evaluation.

In certain neuropathic presentations, the presence of a particular paraprotein or light chain may significantly change prognosis and management. The case of CANOMAD, usually associated with an IgM paraprotein,32 33 40 is discussed previously.

Anti-MAG autoantibodies are the MAG-targeted antibodies of a clonal IgM disorder. The diagnosis of anti-MAG neuropathy is not made simply by finding a positive result for an IgM paraproteinaemic anti-MAG antibody in the serum. The widely used Bühlmann antibody test is very sensitive and, although relatively specific for anti-MAG autoantibodies, is non-specific when applied to a population; anti-MAG autoantibodies are part of an intrinsic repertoire and are very frequently expanded in patients with Waldenström’s macroglobulinaemia. Their presence implies neither that they are pathogenic nor that there is even a neuropathy. Patients with IgM anti-MAG paraproteinaemic neuropathy first have a slowly progressive sensory ataxia, relatively mild weakness, tremor and an IgM (usually kappa) paraprotein. The anti-MAG antibodies are usually ‘strongly positive’ or at least ‘positive’.33 40 There is a characteristic distal, symmetrical demyelinating neurophysiology with disproportionately prolonged distal motor latencies. Biopsies are rarely performed, usually when there is concern about alternative pathologies, but when they are done in anti-MAG neuropathy, the pathogenic changes of widely spaced myelin are characteristic. Anti-MAG titres and IgM paraprotein levels do not relate well to neuropathy severity or treatment response.33 Therefore, anti-MAG antibodies should only be tested when there is an IgM paraprotein and are probably only relevant when they form part of the typical clinical picture. Box 2 illustrates cases where anti-MAG titres were either useful or unhelpful diagnostically.

Box 2

Neuropathies associated with anti-MAG antibodies


Slowly progressive, distal, symmetrical demyelinating neuropathy with sensory ataxia, relatively mild weakness, frequent tremor, IgM (usually) kappa paraprotein.

Case 1

A 63-year-old man gave a 5-year history of ascending numbness, distal clumsiness and unsteadiness on eye closure. Examination and NCS/EMG were in keeping with a distal acquired demyelinating symmetric phenotype, and he had an IgM kappa paraprotein. Anti-MAG titre was strongly positive. He remained stable for a few years but was referred for treatment of his paraprotein due to slow progression.

Learning point 1

Classical anti-MAG neuropathy has a distinct phenotype and may remain stable or slowly progressive for many years.

Cases 2 and 3

A 63-year-old man was referred with several years’ history of progressive gait difficulties and falls and reduced dexterity. Blood tests identified Waldenström’s macroglobulinaemia. He had low positive anti-MAG titres and normal NCS/EMG. Examination identified extrapyramidal changes. He was diagnosed with corticobasal syndrome; his anti-MAG antibody and Waldenström’s macroglobulinaemia were entirely incidental.

A 49-year-old woman was referred with a 3-year history of an ascending sensorimotor neuropathy with uniform conduction slowing on NCS/EMG. Initial treatment with corticosteroids and intravenous immunoglobulin was ineffective. She had a low-grade IgM kappa paraprotein and ‘positive’ anti-MAG titres. The diagnosis of anti-MAG neuropathy was entertained, but the uniformity of conduction slowing and lack of wide-spaced myelin on nerve biopsy led to genetic testing, which confirmed a pathogenic heterozygous mutation in myelin protein zero, in keeping with Charcot-Marie-Tooth disease type 1B.

Learning point 2:

In cases where the clinical presentation is atypical for anti-MAG neuropathy, the presence of anti-MAG antibody (particularly low positives) may be incidental.

  • EMG, electromyography; MAG, myelin associated glycoprotein; NCS, nerve conduction studies.

Other blood biomarkers

In patients with subacute, progressive length-dependent, motor-predominant, often painful neuropathy associated with an IgG or IgA lambda paraprotein, the neurologist should consider POEMS syndrome.41 Over half of patients are initially diagnosed as having CIDP or GBS and are treated with serial ineffective courses of IVIg, corticosteroids or other drugs. In the patient presenting with a subacute or chronic demyelinating neuropathy, a serum immunofixation and urine Bence Jones protein (to identify the low-level lambda light-chain paraprotein) and a serum or plasma vascular endothelial growth factor (VEGF) have almost perfect sensitivity and specificity to identify POEMS and save money and disability when used early.42 43 The immunofixation and the VEGF used here in the context of a demyelinating neuropathy and a lambda light-chain paraprotein form a powerful complex diagnostic biomarker. The other legion features of POEMS can then be sought if not already found.44 45 VEGF may be falsely normal in POEMS patients recently treated with corticosteroids, and VEGF is raised in iron deficiency anaemia, hypoxia-associated conditions (eg, sleep apnoea) or some bone-involving tumours.41 Box 3 illustrates examples where elevated VEGF either helps the diagnosis or contributes to diagnostic confusion.

Box 3

Neuropathies associated with elevated VEGF


Subacute, progressive length-dependent, motor-predominant, often painful neuropathy, IgG or IgA lambda paraprotein, systemic features (papilloedema, organomegaly, endocrine disturbance, skin changes, oedema, etc).

Case 1

A 39-year-old woman had a 2-year history of pain and numbness in her lower limbs, ankle swelling, tiredness, irregular periods and light-headedness on standing. On inspection, there were small glomerular haemangiomata on her chest. She had papillo-oedema and lower limb pitting oedema. NCS/EMG reported a mixed axonal and demyelinating neuropathy. She had an IgA lambda paraprotein and early morning cortisol was low. Serum VEGF was markedly raised. She started on lenalidomide and dexamethasone with a view to stem cell transplantation as definitive treatment.

Learning point 1

POEMS presents with multisystem abnormalities, a typical paraprotein and a distinctive neuropathy. Some features may not be volunteered and should be sought for during evaluation. Raised serum VEGF becomes interpretable in the context of the correct neuropathy and light-chain paraprotein.

Case 2

A 51-year-old man gave an 8-month history of ascending pain and altered sensation, leg swelling, fatigue and weakness. He had an IgG lambda paraprotein. NCS/EMG showed a length-dependent sensory motor mixed neuropathy with axonal changes and slight conduction slowing. VEGF was only slightly raised above upper limit of normal (771 pg/mL). He also had a history of cryoglobulinaemia, raising the query of a vasculitic process. However, 18F-fluorodeoxygluclose (FDG) positron emission tomography/computed tomography (PET-CT) revealed skeletal lesions, and sural nerve biopsy showed uncompacted myelin with no vasculitic changes. He was treated with stem cell transplantation with good improvement.

Learning point 2

VEGF is usually raised in POEMS above 1000 pg/mL (upper limit of normal 771 pg/mL). In a minority of patients, VEGF may be lower, requiring careful testing to establish POEMS diagnosis.

Case 3

A 56-year-old woman with a recent diagnosis of POEMS, for which she had started lenalidomide and dexamethasone in preparation for autologous stem cell transplant, was admitted to the intensive care unit with a severe fungal chest infection. Her serum VEGF, which had normalised with treatment, again rose to abnormal levels. This was attributed to hypoxia from a combination of her infection and methaemoglobulinaemia due to dapsone treatment, rather than relapse, and her VEGF normalised with resolution of her hypoxia.

Learning point 3

VEGF may be falsely raised in hypoxia and should be considered as a cause for rises in VEGF even in confirmed POEMS cases.

  • EMG, electromyography; NCS, nerve conduction studies; POEMS, polyneuropathy, organomegaly, endocrinopathy, monoclonal protein and skin changes syndrome; VEGF, vascular endothelial growth factor.

Painful, rapidly evolving, often asymmetrical plexoradiculoneuropathy should trigger consideration of an infiltrative peripheral nerve process such as neurolymphomatosis.46 Concomitant bladder difficulty or cranial nerve involvement is particularly suggestive, often rapidly responsive to corticosteroids. Low-grade non-Hodgkin’s lymphomas, transformed diffuse large B-cell lymphoma, and less commonly or rarely Waldenström’s macroglobulinaemia (Bing-Neel syndrome) or chronic lymphocytic leukaemia may also present similarly.47 These conditions lack reliable diagnostic blood biomarkers, and CSF studies (discussed further) and a careful search for tissue amenable to biopsy are important.

Abnormalities in serological biomarkers are only significant in the appropriate context. Equally, clinicians should be cautious in excluding a clinically likely diagnosis on the basis of normal blood biomarkers alone. Box 4 illustrates this with examples of systemic and non-systemic vasculitic neuropathy.

Box 4

Neuropathies associated with blood serological markers of systemic vasculitis


Often painful, asymmetrical multiple mononeuropathy with or without systemic features.

Case 1

A 29-year-old woman with a history of childhood asthma, recent sinusitis and itchy erythematous rash presented with asymmetrical painful lower limb neuropathy, elevated inflammatory markers, elevated eosinophil count and raised myeloperoxidase. She was diagnosed with eosinophilic granulomatous polyangiitis and was started on corticosteroids and mycophenolate mofetil.

Learning point 1

Patients with typical systemic vasculitis and neuropathy may be treated for vasculitis without nerve biopsy. Systemic symptoms may precede neuropathy by many years.

Case 2:

A 70-year-old woman gave a 2-year to 3-year history of bilateral ascending leg pain and unsteadiness. Nerve conduction studies/electromyography showed large-fibre axonal changes. Systemic inflammatory markers and cerebrospinal fluid were normal. A sural nerve biopsy showed definite evidence of vasculitis. She was started on immunotherapy with stabilisation of her symptoms.

Learning point 2

In non-systemic peripheral nerve vasculitis, blood serological markers of vasculitis are absent. Biopsy of an affected nerve is required for definitive diagnosis.

CSF testing in neuropathies

The interpretation of CSF findings in neuropathy diagnosis and management is enhanced by the appreciation of the limitations and opportunities offered by CSF analysis.

A raised CSF white cell count in peripheral nerve disease is usually associated with nerve root (radicular) involvement, which can be infiltrative, infective or inflammatory. A CSF pleocytosis (usually <10 cells/µL and certainly <50 cells/µL) occurs in 15% of GBS.48 A greater pleocytosis should trigger further investigation. In acute and subacute radiculoneuropathies, a CSF lymphocytosis is associated with Lyme disease and other tick-borne fever, mycoplasma, sarcoid, HIV, and herpes viruses and enteroviruses. Tuberculosis, listeria and fungi do not cause radicular disease in the absence of other extensive disease. Syphilis does not cause neuropathy. The presence of neutrophils more suggests an infective cause or myelonecrosis. Lymphomatous radiculomeningitis is not uncommon and should be considered in any CSF lymphocytosis.

Albuminocytological dissociation (raised CSF total protein without lymphocytosis) indicates blood–nerve or blood–brain barrier dysfunction. A raised CSF total protein has limited discriminatory value between causes of neuropathy or even non-neuropathic processes. A recent American study found that 50% of cases of ‘CIDP’ were misdiagnosed as such, mostly through undue emphasis on raised CSF protein of <1 g/L.49 Indeed, Allen and Lewis have suggested that there should be a new ‘threshold’ of >1 g/L for even taking the CSF into consideration in CIDP diagnosis.49 Equally, a normal CSF protein does not rule out an inflammatory radiculoneuritis. In CIDP, CSF protein is completely normal in about 6% of patients50 (with many more having <1 g/L), and in GBS, CSF protein is normal in 30%–50% in the first week and 10%–30% even in the second week.15

The CSF ‘total protein’ analysis is limited as it can contain any number of proteins with different meanings. The CSF IgG and albumin give important evidence about blood–nerve barrier dysfunction and inform the oligoclonal band result. IgG monoclonal bands can be identified as a marker of IgG lymphomas.51 CSF IgM is seldom measured but increases in inflammatory and infective processes affecting the meninges. The IgM index (the ratio of the quotients for IgM and albumin ((IgMcsf/IgMserum)/(Albcsf/Albserum)) rises hugely in lymphomatous meningoradiculopathy.51 As described previously, a plexoradiculoneuropathy in conjunction with evidence of B-cell proliferation should prompt consideration of Bing-Neel syndrome, and in this case, CSF testing for the MYD88 mutation may help (box 5). CSF and serum neurofilament light chain (NFL) is raised in many conditions, resulting in axonal damage, and other biomarkers are being developed.

Box 5

Neuropathies associated with CSF MyD88 mutations (‘peripheral’ Bing-Neel syndrome)


Progressive radicular/neuropathic symptoms with evidence of monotypic B-cell proliferation. Central features and cranial neuropathies may be present.

Case 1

A 69-year-old woman attended with gradually worsening balance, distal sensory alteration and proximal and distal mild weakness. She had an IgM kappa paraprotein, and CSF showed pathologically high concentrations of IgM and a positive MYD88 mutation. NCS/EMG showed a severe symmetrical demyelinating sensorimotor neuropathy. Her symptoms stabilised with chemotherapy.

Case 2

A 43-year-old man presented with constitutional symptoms, IgM kappa paraprotein and lymphoplasmacytoid cells in the bone marrow. He was diagnosed with and treated for Waldenström’s macroglobulinaemia. Ten months later, he developed rapidly progressive, symmetric sensorimotor deficits. NCS/EMG showed non-length-dependent axonal neuropathy with patchy conduction slowing. CSF MYD88 was negative. However, a sural nerve biopsy showed a patchy but prominent Waldenström’s macroglobulinaemia cell infiltrate. Further treatment for this condition markedly improved neuropathic symptoms.

Learning points

CSF MYD88 L265P mutations are present in most patients with Bing-Neel syndrome. However, in the extremely rare purely peripheral cases of Bing-Neel syndrome, biopsy of affected nerves may be required to identify pathological Waldenström’s macroglobulinaemia infiltration.

  • CSF, cerebrospinal fluid; EMG, electromyography; NCS, nerve conduction studies.

Imaging biomarkers in peripheral neuropathy

Imaging of peripheral nerves with MRI and ultrasound scanning has been explored for over 20 years52 53 and has entered diagnostic guidelines.54 55 Nerve thickening and signal abnormality are present in a variety of genetic56 and acquired57 58 neuropathies, including CIDP57 and acute inflammatory demyelinating polyneuropathy,58 MMNCB,59 POEMS60 and neurolymphomatosis,61 infective neuropathies (especially leprosy)62 and rarely sarcoidosis.63

Patterns of enlargement, smooth or focal swellings and gadolinium enhancement provide limited discrimination between one neuropathy and another. Healthy normal values, even in the largest nerves, vary widely, and in disease, significant nerve-to-nerve variations exist.64 Current MRI resolution limits assessable targets to the brachial and lumbosacral plexus and the major peripheral nerves in the limbs, and experienced ultrasound technicians are few with low inter-rater reliability.65

Figure 2 shows optimised plexus MRI in various diseases showing that MRI has limited discriminatory ability in diffuse neuropathies. Even though MR neurography66 and ultrasound scanning67 are increasingly recognised as helpful in establishing diagnosis of such neuropathies, current modalities tend to confirm prior expectations, limiting their real-world diagnostic utility. However, in these scenarios, peripheral nerve imaging can be helpful in localising rather than defining pathology, especially for targeting nerve biopsy (figure 3).

Figure 2

Peripheral nerve MRI showing thickening and signal abnormality of the brachial and lumbosacral plexus in various pathologies. Coronal STIR and fat-suppressed postcontrast T1-weighted images of brachial (A,B) and lumbosacral (C,D) plexi in two patients with CIDP showing thickening, STIR hyperintensity and faint enhancement of the proximal components of the plexi, with relative symmetry in one and conspicuous asymmetry in the second. Coronal and sagittal oblique STIR images of the brachial plexus (E,F) showing symmetric signal abnormality and thickening but no enhancement (not shown) in suspected hereditary hypertrophic neuropathy. Similar but more patchy signal change and brachial plexus thickening in paraprotein-related neuropathy (G). Massive lumbosacral plexus and proximal sciatic nerve hypertrophy (H) in probable CIDP, initially suspected to be hereditary neuropathy. CIDP, chronic inflammatory demyelinating polyradiculoneuropathy; STIR, short tau inversion recovery.

Figure 3

Utility of peripheral nerve MRI in the evaluation of neuropathy. Coronal fat-suppressed postcontrast T1w image of the thighs, showing typical fusiform enlargement and enhancement of left sciatic perineurioma (A), with a further axial fat-suppressed postcontrast T1w image of the right thigh (B) showing preserved fascicular architecture (arrow) of a right proximal common peroneal perineurioma. Sagittal T1w and axial STIR images of the left upper arm demonstrating classical ‘target’ appearance of a large, well-defined left radial nerve schwannoma (C,D). Coronal pelvic STIR identifying extensive well-circumscribed nodular lesions extending along the lumbosacral plexus, pelvic nerves and left sciatic nerve (dashed arrows) in neurofibromatosis type 1 (E). Coronal STIR and fat-suppressed postcontrast T1w showing nodular thickening and enhancement of the right brachial plexus (arrowheads) and adjacent supraclavicular lymphadenopathy (curved arrow) consistent with metastatic infiltrative breast carcinoma (F,G). The value of imaging in lesion localisation: sagittal T2w lumbar spine image (H) showing no evidence of neural compression in a patient presenting with clinical features of a right S1 radiculopathy, and subsequent peripheral nerve MRI (I) identifying a focal enhancing lesion of the right proximal tibial nerve in the thigh (arrow). A coronal fat-suppressed postcontrast T1w image (J) shows an enhancing lesion involving the right brachial plexus posterior cord (arrowheads) in relapsed diffuse large B-cell lymphoma, used to guide a targeted biopsy to confirm the diagnosis. STIR, short tau inversion recovery; T1w, T1-weighted; T2w, T2-weighted.

MR neurography has superior diagnostic accuracy (94% vs 80% for ultrasound scanning) for focal pathologies, but ultrasound scanning is quicker and more cost effective.68 Figure 3 illustrates situations where peripheral nerve MRI is useful in clinical management. MRI can localise abnormalities suitable for biopsy in a peripheral nerve. The imaging appearances can organise a differential diagnosis while awaiting pathological confirmation. This is particularly useful where electrophysiological localisation of the lesion is difficult, such as in the radial nerve. Imaging may also help in revealing the extent of disease burden and in the assessment of progression or response to treatment.

Muscle MRI not only gives diagnostic information in primary muscle disease but can help in neuropathy. Intramuscular fat accumulation is the final tissue outcome for myopathies and neuropathies because of downstream muscle denervation. Figure 4 shows progressive abnormalities over time and disease progression of muscle fat fraction in Charcot-Marie-Tooth (CMT) type 1A by three-point Dixon quantification. This technique has superior responsiveness to change compared with existing Charcot–Marie–Tooth clinical outcome scores.69 Further studies are investigating MRI with novel sequences to use as biomarkers of disease activity in other neuropathies.

Figure 4

Axial MR images of the right lower leg muscles in CMT1A showing progression of intramuscular fat accumulation. (A) T1-weighted image shows hyperintensity in the anterolateral compartment and both heads of gastrocnemius but does not allow precise quantification. (B–D) Three-point Dixon FF maps (greyscale: black 0% fat; white 100% fat). (B) Baseline FF map with muscle segmentation (red: anterior compartment, green: peroneus longus, blue: lateral head of gastrocnemius, yellow: medial head of gastrocnemius, cyan: soleus, magenta: tibialis posterior). (C) Baseline FF map without segmentation: mean FF within calf muscles is 23.9%. (D) FF map at 12-month follow-up. Visually there is significant progression in the anterolateral compartment; on quantification, mean FF has increased to 27.7%. FF, fat fraction.

Clinical outcome scores as neuropathy biomarkers

Despite huge advances in biochemical, immunological and imaging technologies and discovery, following the course of a neuropathy and its response to interventions remains largely clinical. There are many clinical outcome scores in common use such as the MRC-SS, the Overall Neuropathy Limitations Score and the INCAT sensory symptom score. These have numerous limitations; the peripheral neuropathy outcome measures standardisation study for inflammatory neuropathies group have succeeded in generating clinical outcome measures of genuine utility.

A working knowledge of strengths and limitations of these measures allows proper interpretation. Impairment measures (eg, grip strength, MRC-SS) can be applied to many neuropathies. However, the MRC-SS is non-linear and statistically invalid for averaging.70 The MRC-SS depends on rater experience and is less helpful in conditions with focal weakness outside the muscle groups tested in the sum score. Disability patient-reported outcome measures have meaning for patients. Modern clinimetric theory has built statistically valid, responsive, disease-specific scales such as the Rasch-built Overall Disability Scales for CIDP, GBS and MMNCB reflecting different disability patterns.71 72 Some scores, such as the second version CMT Neuropathy Score (CMTNS2), are composite and include symptoms, examination findings and electrophysiological variables73; these fail in being statistically invalid to average, non-discriminating because of the dilutional factor of multiple elements, and relatively non-responsive to change.

With some neuropathies (such as CIDP) being particularly heterogeneous, it is difficult to identify a single outcome measure that has universal relevance. Concomitant non-neuropathic physical disability, mood disturbance and functional overlay may also affect performance.

Whatever score is used, the minimal clinically important difference (MCID), established through a mixture of statistics and expert opinion, is important to appreciate. Its derivation and actual value for any one score is complex, but it delineates an important threshold of change.74

In clinical practice, disease activity should be measured with a combination of multiple outcome measures. Published recommendations for assessment may not appreciate this. For instance, NHS commissioning guidelines for IVIg use requires demonstration of improvement in disability in at least one of three outcome measures judged by the clinician to be appropriate for the patient’s disability.75 In fact, contiguous small (sub-MCID) changes in more than one outcome measure may be more specific and sensitive at identifying relevant change than larger single outcome measure changes.76

Future directions in neuropathy biomarker development

Single molecular assay platforms such as SiMoA and OLink explore biomarker detection at ultralow levels and across multiple candidates.77 NFL is a now well-known but non-specific biomarker of nerve degeneration in the central and peripheral nervous systems.78 Plasma NFL has been explored as a biomarker for CMT disease,79 hereditary amyloidosis,80 CIDP81 and GBS.82 NFL is not specific to peripheral nerve damage and only really reflects axonal damage. We still lack serological biomarkers that are more specific to the peripheral nervous system and sensitive to demyelinating pathologies.


A range of biomarkers are available to the neurologist to aid diagnosis and management. We must be aware of the accuracy and limitations of tests in different clinical scenarios, and investigations should be requested with careful consideration of how they are interpreted. Future research may allow for development of sensitive, specific, responsive serological and imaging biomarkers which allow easier differentiation of different types of neuropathy and provide objective means of monitoring disease activity.

Key points

  • Clinical context is key in managing peripheral neuropathies; knowing when to disregard an unremarkable positive test result is important in avoiding misdiagnosis.

  • Certain blood biomarkers such as myelin associated glycoprotein autoantibodies and vascular endothelial growth factor are interpretable only in the context of a compatible neuropathy and the presence of an appropriate paraprotein.

  • Imaging may help to localise a peripheral neuropathy—particularly for targeting nerve biopsy—and may help to monitor disease activity.

  • For most neuropathies, clinical outcome measures are better than investigation results in monitoring disease activity and treatment response.

Further reading

  • Fuller G. How to get the most out of nerve conduction studies and electromyography. J Neurol Neurosurg Psychiatry 2005;76(Suppl 2):ii41–46. doi: 10.1136/jnnp.2005.067355

  • Delmont E, Willison H. Diagnostic Utility of Auto Antibodies in Inflammatory Nerve Disorders. J Neuromuscul Dis 2015;2 (2):107–112. doi:10.3233/JND-150078

Data availability statement

Data sharing not applicable as no datasets generated and/or analysed for this study.

Ethics statements

Patient consent for publication

Ethics approval

Not applicable.



  • Twitter @mike_the_nerve

  • Collaborators N/A.

  • Contributors All authors were involved in conception, drafting and revision of the manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests No specific funding was received for this work. MPL, ASC and JM are supported by the National Institute for Health Research, University College London Hospitals NHS Foundation Trust and Biomedical Research Centre. RYSK is funded by GBS‐CIDP Foundation International.

  • Provenance and peer review Commissioned; externally peer reviewed by Simon Rinaldi, Oxford, UK.

Linked Articles

  • Highlights from this issue
    Phil E M Smith Geraint N Fuller

Other content recommended for you