Intended for healthcare professionals

Clinical Review State of the Art Review

Diagnosis and management of sensory polyneuropathy

BMJ 2019; 365 doi: (Published 08 May 2019) Cite this as: BMJ 2019;365:l1108
  1. Kelly Graham Gwathmey, assistant professor of neurology1,
  2. Kathleen T Pearson, assistant professor of neurology1
  1. 1Virginia Commonwealth University, Department of Neurology, 1101 E. Marshall Street, PO Box 980599, Richmond, VA 23298, USA
  1. Correspondence to: K G Gwathmey Kelly.Gwathmey{at}


Sensory polyneuropathies, which are caused by dysfunction of peripheral sensory nerve fibers, are a heterogeneous group of disorders that range from the common diabetic neuropathy to the rare sensory neuronopathies. The presenting symptoms, acuity, time course, severity, and subsequent morbidity vary and depend on the type of fiber that is affected and the underlying cause. Damage to small thinly myelinated and unmyelinated nerve fibers results in neuropathic pain, whereas damage to large myelinated sensory afferents results in proprioceptive deficits and ataxia. The causes of these disorders are diverse and include metabolic, toxic, infectious, inflammatory, autoimmune, and genetic conditions. Idiopathic sensory polyneuropathies are common although they should be considered a diagnosis of exclusion. The diagnostic evaluation involves electrophysiologic testing including nerve conduction studies, histopathologic analysis of nerve tissue, serum studies, and sometimes autonomic testing and cerebrospinal fluid analysis. The treatment of these diseases depends on the underlying cause and may include immunotherapy, mitigation of risk factors, symptomatic treatment, and gene therapy, such as the recently developed RNA interference and antisense oligonucleotide therapies for transthyretin familial amyloid polyneuropathy. Many of these disorders have no directed treatment, in which case management remains symptomatic and supportive. More research is needed into the underlying pathophysiology of nerve damage in these polyneuropathies to guide advances in treatment.


Peripheral sensory nerves vary in size and function, ranging from the smallest unmyelinated C fibers and thinly myelinated Aδ fibers that conduct noxious and thermal information12 to the larger Aβ fibers that transmit proprioceptive and vibratory information.3 As a result, disorders of sensory nerve function are diverse and depend on the type of nerve fiber that is affected; patients present with a wide range of symptoms, from pain predominant (small fiber) to ataxia predominant (large fiber) problems. This article will not attempt to review all peripheral sensory neuropathies that manifest the classic length dependent or “stocking glove” pattern, but will focus on those that have a clearly pain predominant or ataxia predominant presentation. It will also include other disorders that present with sensory ataxia but affect the dorsal root ganglia (DRG), sensory fibers of the nerve roots, and dorsal columns. We will also cover the differential diagnosis of sensory polyneuropathies, the diagnostic approach to patients with sensory problems, and disease specific and symptomatic treatments.

Sources and selection criteria

We searched PubMed for English language articles published from 1 January 2000 to 1 October 2018 using the terms “small fiber neuropathy”, “sensory ataxia”, “sensory neuronopathy”, “dorsal root ganglionopathy”, “dorsal root ganglion”, “skin biopsy”, “quantitative sensory testing”, “corneal confocal microscopy”, “quantitative sudomotor axon reflex testing”, “thermoregulatory sweat testing”, “electrochemical skin conductance”, “sarcoidosis”, “Sjögren’s syndrome”, “fibromyalgia”, “sodium channelopathies”, “transthyretin”, “sensory Guillain-Barré syndrome”, “ataxic Guillain-Barré syndrome”, “acute sensory ataxic neuropathy”, “Miller Fisher syndrome”, “disialosyl antibodies”, “ganglioside antibodies”, “CANOMAD”, “CANDA”, “sensory chronic inflammatory demyelinating polyneuropathy”, “distal acquired demyelinating symmetric neuropathy”, “anti-MAG”, “anti-Hu”, and “tabes dorsalis”. We included a few articles of historical importance that were published in the 1980s and 1990s. These sentinel articles set the conceptual framework for these disorders and their inclusion was necessary. We searched reference lists of articles selected through title, abstract, and full text review. We selected randomized controlled trials, observational, and basic science studies, systematic reviews, and meta-analyses from these sources. Articles were prioritized by study quality and topic. Given that many of the sensory neuropathies discussed are extremely rare, case studies and case series were also reviewed and included if deemed important.

List of acronyms

  • AAN: American Academy of Neurology

  • ANA: Antinuclear antibodies

  • ASAN: Acute sensory ataxic neuropathy

  • BPI-MSF: Brief Pain Inventory Modified Short Form

  • CANDA: Chronic ataxic neuropathy with disialosyl antibodies

  • CANOMAD: Chronic ataxic neuropathy, ophthalmoplegia, IgM paraprotein, cold agglutinins, and disialosyl antibodies

  • CIDP: Chronic inflammatory demyelinating polyneuropathy

  • CISP: Chronic immune sensory polyradiculopathy

  • DADS: Distal acquired demyelinating symmetric neuropathy

  • DRG: Dorsal root ganglia

  • EFNS: European Federation of Neurological Societies

  • EMLA: Eutectic mixture of local anesthetic

  • ESR: Erythrocyte sedimentation rate

  • GBS: Guillain-Barré syndrome

  • IENFD: Intraepidermal nerve fiber density

  • IFG: Impaired fasting glucose

  • IGT: Impaired glucose tolerance

  • LEP: Laser evoked potential

  • MAG: Myelin associated glycoprotein

  • MFS: Miller-Fisher syndrome

  • MRI: Magnetic resonance imaging

  • mNIS+7: Modified Neuropathy Impairment Score +7

  • Norfolk QOL-DN: Norfolk Quality of Life-Diabetic Neuropathy

  • NPS: Neuropathic Pain Scale

  • QSART: Quantitative sudomotor axon reflex test

  • SFN: Small fiber neuropathy

  • SGPG: Sulphated glucuronyl paragloboside

  • SNAP: Sensory nerve action potential

  • SNRI: Serotonin-norepinephrine reuptake inhibitor

  • SSEP: Somatosensory evoked potential

  • TCA: Tricyclic antidepressant

  • TTR-FAP: Transthyretin familial amyloidosis with polyneuropathy

Incidence and prevalence

The sensory polyneuropathy category includes extremely common conditions such as diabetic neuropathies (the most common cause of neuropathy worldwide) and very rare conditions, such as specific acute ataxic neuropathies (described only in case series). Table 1 lists the incidence and prevalence of these specific polyneuropathies and their underlying causes, if known.

Table 1

Incidence and prevalence of sensory neuropathies highlighted in this review*

View this table:

Clinical presentation

The clinical presentation and findings on physical examination depend on the type of affected nerve fiber and the distribution of nerve damage. Patients may report a combination of positive (paresthesia, burning pain) and negative (loss of sensation) sensory disturbances, as well as gait imbalance. Important considerations regarding the clinical presentation include acuity of onset, time course of progression, and the distribution and quality of sensory symptoms.

Small fiber neuropathies

In small fiber neuropathies (SFNs) the thinly myelinated (Aδ) and unmyelinated (C) fibers responsible for the transmission of thermal and noxious sensory input are affected.12 Clinically, this nerve damage translates to symptoms of sharp, painful, or burning paresthesia; sensory loss or numbness; and the inability to discriminate between hot and cold sensations. Symptoms may be vague, described as a tight feeling or abnormal sensation in the soles of the feet, intolerance of tactile stimuli (inability to wear socks or touch bedsheets), or a sensation of restless legs. The distribution of symptoms may have a length dependent or non-length dependent pattern that affects the limbs, trunk, face, or it may have a combination of patterns.12404142 Depending on the underlying cause, the onset of symptoms may be gradual, with slowly progressive worsening, or subacute with more rapid progression. Pain may be prominent and disabling, and a recent large Italian cohort study of patients with painful diabetic neuropathy suggests that pain may be more common in women.43

Dysautonomia is often a feature of SFN owing to impairment of the sympathetic and parasympathetic function of Aδ fibers and the postganglionic autonomic function of C fibers. It is essential to ask patients about potential autonomic involvement including orthostasis; palpitations; abnormal sweating; dry mouth, eyes, or skin; gastrointestinal symptoms including cramping, diarrhea, or constipation; flushing or other changes of skin color; and erectile dysfunction.2

A patient with SFN may have decreased temperature and pinprick sensation on examination, and potentially allodynia, dysesthesia, or hyperesthesia on sensory testing. Motor strength, proprioception, and muscle stretch reflexes should be preserved in patients with pure SFN. Skin may have a dry, atrophic, or discolored appearance.1240

Sensory ataxia

Disorders affecting the large myelinated Aβ fibers, 1a fibers, sensory nerve roots, or DRG will result in impaired vibratory sensation and proprioception. Clinically this results in a combination of symptoms of sensory loss, paresthesia, and gait imbalance. The ataxic sensory polyneuropathies will present acutely or have an insidious onset and gradually progressive course as a result of dysfunction of the peripheral sensory nerves. Physical examination may show absent or reduced vibratory sensations, abnormal proprioception, depressed or absent reflexes, and sensory ataxia.

In sensory neuronopathies (dorsal root ganglionopathies), sensory neurons of the dorsal root and trigeminal ganglia are affected. The clinical presentation is characterized by pronounced ataxia and sensory loss, which may have a non-length dependent or multifocal pattern. In addition, pain and positive sensory symptoms often occur because of the involvement of small and medium sized nerve fibers.44 The face and trunk may also be affected.444546 The results of a physical examination will resemble that seen in patients with ataxic sensory polyneuropathies, although the sensory deficits are more often patchy, non-length dependent, or generalized. The finding of pseudoathetosis, as a result of impaired afferent proprioceptive input, is a hallmark of DRG dysfunction.444647 Although motor strength is preserved in pure sensory neuronopathies, it may seem to be impaired on examination owing to the lack of proprioceptive input during confrontational strength testing. The clinical course may be gradual and insidious in idiopathic forms of the disease, but it will typically have a subacute course in patients with paraneoplastic, immune mediated, and toxic causes.44

Patients with dorsal column dysfunction may also present with sensory ataxia. Often these patients also have evidence of upper motor neuron signs on examination, which suggests corticospinal tract involvement and will guide the examiner away from localization in the peripheral nervous system. When the dorsal columns and corticospinal tracts are affected, patients will have spasticity, weakness, and reduced vibratory and proprioceptive sensations: the so called posterolateral column syndrome.48

Differential diagnosis of small fiber neuropathies

The causes of SFN fall into six broad categories: metabolic, inflammatory, genetic, toxic, infectious, and idiopathic (cryptogenic) (table 2). Many of the known common causes will not be discussed in detail but are included in table 2. Fibromyalgia, which has been associated with pathologic evidence of SFN, does not easily fall into one of the six categories. Alternatively, classification based on clinical phenotype has also been proposed.49 Despite extensive evaluation, 20-50% of cases of SFN are ultimately classified as idiopathic.50515253 The most common causes include diabetes, immunologic conditions, sodium channel mutations, and vitamin B12 deficiency.29 Although immunologic conditions were found in 19% of a cohort of 921 patients with SFN, which exceeds the prevalence in the general population, the exact pathogenic role of isolated autoantibodies remains unclear.2954 In one series, the highest yield blood tests in SFN that appeared to be “initially idiopathic” were erythrocyte sedimentation rate (ESR), antinuclear antibodies (ANA), C3 complement values, and autoantibodies that are associated with Sjögren’s syndrome and celiac disease.55 It has been recommended that patients are screened for glucose intolerance, vitamin B12 deficiency, and sodium channel mutations even if there is a known underlying cause.2954

Table 2

Causes of small fiber neuropathy and ancillary investigations*

View this table:

Metabolic causes: diabetes and prediabetes

Diabetes is the most common cause of polyneuropathy worldwide and the most common cause of SFN specifically.56 The association between prediabetes (impaired glucose tolerance (IGT) and impaired fasting glucose (IFG)) and polyneuropathy is still being delineated. IGT is defined by a raised two hour glucose level on an oral glucose tolerance test of 7.8-11.1 mmol/L (140-199 mg/dL). IFG is defined by a fasting glucose of 5.6-6.9 mmol/L (100-125 mg/dL). It is likely that the risk of neuropathy is higher for IGT than for IFG.57 When considering the diagnostic investigations in these patients, it is important to note that glycosylated hemoglobin may be normal in patients with IGT.58

Some studies support an association between IGT and polyneuropathies,1059606162 whereas others have failed to show such a correlation.636465 It is thought that IGT associated neuropathy mainly affects the small nerve fibers, perhaps explaining why some researchers have found no correlation between IGT and large fiber polyneuropathy60666768 and others have questioned the association between IGT and SFN.6469 Such incongruent findings across studies are probably the result of differences in definitions of polyneuropathy (including the use of symptoms or intraepidermal nerve fiber density (IENFD)), degrees of surveillance, and polyneuropathy endpoints.69 Nonetheless, the identification of prediabetes is of utmost importance because 50% of patients with prediabetes ultimately develop type 2 diabetes,70 and reducing the risk of conversion to diabetes decreases the risk of developing polyneuropathy.

The Impaired Glucose Tolerance Neuropathy study investigated 32 patients with IGT and neuropathy. It found that 65% of patients had low amplitude or absent sural responses, 83% had decreased IENFD, and 61% had abnormal quantitative sudomotor autonomic reflex test results.71 Skin biopsy was found to be the most sensitive measure of the severity of IGT related neuropathy, and partial cutaneous reinnervation was seen after the introduction of a suitable diet and exercise. Other features of the metabolic syndrome, including hypertriglyceridemia and central obesity, are also independent risk factors for SFN.72

Autoimmune causes

The known autoimmune causes of SFN are diverse and include sarcoidosis and Sjögren’s syndrome in addition to systemic lupus erythematosus, celiac disease, and others.


SFN is the most common peripheral nervous system manifestation in sarcoidosis, and its pathophysiology is probably related to a systemic release of inflammatory mediators rather than granulomatous involvement of the small nerve fibers.16177374 Unlike pulmonary sarcoidosis, which preferentially affects African-Americans, SFN seems to affects mainly white people.75 Most patients will have a non-length dependent pattern of numbness, pain, and paresthesia. Half will develop dysautonomia, with orthostasis being the most common manifestation.75

Sjögren’s syndrome

SFN is probably the most common neuropathic manifestation of Sjögren’s syndrome.7677 The onset of symptoms is subacute to chronic (weeks to months) although hyperacute presentations have been reported.7778 Serologic testing is often unhelpful—the estimated sensitivities of anti-SSA (anti-Ro) and anti-SSB (anti-La) antibodies are 39% of 17%, respectively.79

Other autoimmune small fiber neuropathies

Some experts have proposed an additional category of “apparently autoimmune” SFN that could account for some forms of otherwise idiopathic SFN.80 Patients in this category, who have evidence of systemic autoimmune disorders and blood markers of autoimmunity, have been described as having an atypical, painful SFN that responds to corticosteroids and intravenous immunoglobulins.55818283 This classification is not universally accepted and these findings need to be reproduced in large prospective clinical trials. Acute onset of painful SFN, which might fall into the Guillain-Barré syndrome (GBS) spectrum, has also recently been described.84

Genetic causes

Two familial causes of SFN—sodium channel mutations and transthyretin familial amyloidosis with polyneuropathy (TTR-FAP)—stand out given recent developments in the understanding of their underlying pathophysiology and the emergence of new treatment modalities.

Sodium channelopathies

The SCN9A, SCN10A, and SCN11A genes encode the Nav1.7, Nav1.8, and Nav1.9 sodium channels, respectively. Mutations in these genes have been described in painful, predominantly SFNs.858687 These mutations produce a gain of function change that results in hyperactive pain signaling in the DRG neurons.88

Transthyretin familial amyloidosis polyneuropathy (TTR-FAP)

TTR-FAP is endemic in Japan, Sweden, Portugal, and Brazil. In Europe and Latin America, the ATTR-Val30Met mutation predominates, whereas the ATTR-Val122Ile mutation is most common in the United States.89 More than 120 TTR gene mutations have been reported to cause amyloidosis.90 These mutations induce transthyretin misfolding and systemic deposition of amyloid, resulting in autosomal dominantly inherited transthyretin amyloidosis. As amyloid progressively accumulates, it leads to multiorgan dysfunction and ultimately death. The first stage of TTR-FAP is a length dependent, small fiber predominant sensory polyneuropathy with autonomic dysfunction. Patients develop progressive difficulty with walking and ultimately cardiomyopathy. The diagnosis is confirmed by DNA testing and the demonstration of amyloid deposits on biopsy.91 In addition, diagnostic tools such as magnetic resonance neurography and radionucleotide cardiac scintigraphy are emerging.89

Other small fiber neuropathies


The association between fibromyalgia syndrome—characterized by chronic widespread pain, fatigue, exercise intolerance, and cognitive problems—and small fiber pathology was first described in 2013.252692 Nearly half of patients with fibromyalgia have evidence of reduced IENFD on skin biopsy, and emerging evidence indicates that nearly a third of patients have a distal large fiber neuropathy as indicated by low medial plantar responses.93 It is unclear whether patients who have fibromyalgia with and without small fiber pathology are clinically distinguishable,94 although some researchers report that paresthesia and autonomic involvement may predict the presence of small fiber dysfunction.95 One prospective study compared 30 patients with fibromyalgia with 34 age and sex matched healthy controls in terms of clinical examination, quantitative sensory testing, skin biopsy, blood and cutaneous miRNA isolation. It found that 51 miRNAs were aberrantly expressed in the white blood cells and miR-let-7d correlated with reduced IEFND in the patients with fibromyalgia. In addition, in one group of patients with fibromyalgia, aberrantly expressed miR-let-7d microRNA in white blood cells correlated with reduced IENFD. In the skin of these patients, miR-let-7d and the downstream target of the insulin-like growth factor-1 receptor were also aberrantly expressed in those with small fiber dysfunction.96

Although the association between small fiber disease and fibromyalgia sheds light on the underlying pathomechanisms of fibromyalgia, most patients with fibromyalgia do not have the typical symptoms of SFN.25 That said, the identification of the presence of small fiber dysfunction in fibromyalgia enables screening for other causes of SFN, such as diabetes.939597

Differential diagnosis of sensory ataxia

The ataxic sensory disorders can be classified on the basis of localization (nerve, nerve root, DRG, dorsal column) and further differentiated by time course (acute, subacute, chronic) (table 3). Although dorsal column disorders are not a peripheral nervous system process, they can mimic ataxic neuropathies and will be briefly discussed. The sensory ataxic disorders will be organized on the basis of localization, cause, and time course.

Table 3

Causes of sensory ataxia and ancillary investigations*

View this table:

Acute inflammatory sensory neuropathies

Sensory Guillain-Barré syndrome

The acute sensory polyneuropathies consist of overlapping clinical phenotypes, and the lines are often blurred between sensory GBS, ataxic GBS, acute sensory ataxic neuropathy (ASAN), and Miller-Fisher syndrome (MFS). In 1981, Asbury proposed diagnostic criteria for sensory GBS that included a monophasic episode of acute onset, diffuse, symmetric sensory symptoms; demyelinating electrodiagnostic features (often apparent on motor studies); and albuminocytologic dissociation.98 Given the scarcity of such reports in the literature, the existence of sensory GBS has been called into question.99

A case series in 2001 reported eight additional patients who met the clinical criteria for sensory GBS.100 Serum autoantibodies (MAG (myelin associated glycoprotein)), GM1, GQ1b, GD1b, anti-Hu, and sulphated glucuronyl paragloboside (SGPG) were normal in the four patients tested. Sensory GBS, owing to its demyelinating features and the absence of ganglioside antibodies, remains separate from the following disorders which share many clinical, electrophysiologic, and laboratory features. These diseases, also classified as GBS variants, are best subdivided into complete MFS and incomplete MFS, which includes the acute ataxic neuropathies (ASAN and ataxic GBS).

Miller-Fisher syndrome

MFS is characterized by a classic clinical triad of ophthalmoplegia, ataxia, and areflexia.101102 Less common clinical features include other cranial neuropathies, blepharoptosis, limb dysesthesia, and micturition problems. The ataxia of MFS is thought to be caused by both impaired proprioception (reversible conduction failure in 1a afferents) and cerebellar dysfunction.103104 As in other forms of GBS, neurologic symptoms often follow an antecedent illness such as infection with Campylobacter jejuni or Haemophilus influenzae.105 The distinctive anti-GQ1b ganglioside antibodies crossreact with surface epitopes of C jejuni, supporting the theory of molecular mimicry between nerve and bacteria.106107 These antibodies also crossreact heavily with ganglioside GT1a.108 Electrodiagnostic studies, in contrast to sensory GBS, show a sensory predominant axonopathy.109 Recovery is gradual but often complete.

Acute ataxic neuropathies

The remaining acute ataxic neuropathies, including both ASAN and ataxic GBS, have recently been classified as incomplete forms of MFS by some experts.110111 In the past, ASAN was not considered to be a GBS variant because affected patients do not meet the diagnostic criteria for sensory GBS and lack demyelinating features on electrodiagnostic studies. Both ASAN and ataxic GBS, however, share many features with MFS including acute ataxia, areflexia, antecedent infection, and antiganglioside antibodies but lack the typical ophthalmoplegia.110112 The presence of a Romberg sign helps differentiate ASAN from ataxic GBS. Patients with ASAN may harbor anti-disialosyl antibodies to GD1b alone or in combination with antibodies to CD3, GQ1b, or GT1a. Autoantibodies against gangliosides without disialosyl epitopes (GD1a and GM3) may also be present.112 Given that patients with ASAN typically have an antecedent infection, monophasic course, and excellent recovery, they should be considered under the rubric of GBS, in the subcategory of acute ataxic neuropathy.112 Ataxic GBS is distinguished by cerebellar-like ataxia and absence of a Romberg sign.113 Similar to MFS, these patients also harbor anti-GQ1b IgG antibodies.114 A retrospective chart review identified 54 patients with acute ataxic neuropathy without ophthalmoplegia. The Romberg sign was absent in 37 patients, who were considered to have ataxic GBS. In the other 17 patients, the Romberg sign was present, consistent with a diagnosis of ASAN. In the 37 patients with ataxic GBS, 24 were GQ1b positive compared with three of the 17 patients with ASAN (P=0.0034). IgG antibodies against GD1b but not GQ1b were more common in patients with ASAN (6/17) than in those with ataxic GBS (5/37), but this did not meet statistical significance (P=0.72).115 However, the opposite was true a minority of the time, suggesting that these diseases lie on a spectrum.

Chronic inflammatory sensory neuropathies

Chronic ataxic neuropathy with disialosyl antibodies (CANDA)

These very rare, acute, and chronic ataxic neuropathies with anti-disialosyl antibodies probably share a common pathogenic mechanism, which is disruption at the node of Ranvier on sensory fibers. Like the acute ataxic neuropathies and MFS, the chronic ataxic neuropathies are also associated with anti-disialosyl antibodies (such as GD1b and GQ1b.) These disialosyl antibody mediated neuropathies can be separately categorized as nodo-paranodopathies.107111116 When the full spectrum of clinical features is present in these disialosyl antibody mediated chronic ataxic neuropathies, the disorder goes by the acronym CANOMAD (chronic ataxic neuropathy, ophthalmoplegia, IgM paraprotein, cold agglutinins, and disialosyl antibodies). CANDA (chronic ataxic neuropathy with disialosyl antibodies) is a more general term and allows for the inclusion of patients without ophthalmoplegia and those in whom the cold agglutinins are IgM antibodies.111 CANDA can relapse, remit, and have cranial neuropathies that result in bulbar dysfunction.117 The disease process in CANDA may be the result of antibody mediated attack of the nerve root, DRG, and nerves.111 In electrophysiologic studies, patients with CANDA have absent or reduced sensory responses and diminished motor responses, including demyelinating features.118119

Sensory chronic inflammatory demyelinating polyneuropathy (CIDP)

Patients with sensory CIDP present with a pure sensory neuropathy with intact strength despite often having evidence of acquired demyelination on motor nerve conduction studies.120121122123 A minority of patients with sensory CIDP probably have electrophysiologic abnormalities in the sensory nerves only.124 Features that differentiate patients with sensory CIDP from those with chronic idiopathic axonal polyneuropathies include early gait ataxia, cranial neuropathy, diffuse hyporeflexia, onset before 55 years of age, and early involvement of the upper extremities.123

A small subset of patients with sensory CIDP have chronic immune sensory polyradiculopathy (CISP) in which the disease is localized to the nerve roots. These patients will have normal routine nerve conduction studies, abnormal somatosensory evoked potentials, raised concentrations of cerebral spinal fluid protein, and enlarged nerve roots on magnetic resonance imaging (MRI), which demonstrate inflammation on biopsy.125

Distal acquired demyelinating symmetric neuropathy (DADS)

Distal acquired demyelinating symmetric neuropathy (DADS), a variant of CIDP, is characterized by distal, symmetric, sensory, or sensorimotor polyneuropathy occurring in the presence of an IgM monoclonal gammopathy and myelin associated glycoprotein (MAG) antibodies.126 Patients who have an identical clinical and electrophysiologic phenotype but lack MAG antibodies can be classified as having DADS-CIDP36127; such patients may carry a better prognosis and respond more favorably to intravenous immunoglobulins, corticosteroids, and plasma exchange.127 The clinical hallmark of DADS neuropathy is the gradual onset of sensory ataxia resulting from impaired proprioception.128 Weakness is less prominent and, when present, affects the distal lower extremities.129 Action tremor can be a prominent feature.130131 The electrophysiologic features include extremely prolonged distal motor and sensory latencies representing distal demyelination.132133 Pathologically, there is segmental demyelination with IgM and complement deposits in the myelin sheaths and widened outer myelin lamellae.134 More than half of patients with DADS have IgM paraproteins that recognize MAG or SGPG (which is present in most patients with anti-MAG antibodies). Three quarters of patients with non-anti-MAG DADS have anti-ganglioside antibodies (GD1b, GQ1b, GT1b, and others).128

Sensory neuronopathies

The sensory neuronopathies, or dorsal root ganglionopathies, are a small subset of sensory polyneuropathies that result from damage to the trigeminal ganglion sensory neurons and DRG. These uncommon disorders can be broadly classified as inherited, autoimmune, or acquired. Because a comprehensive discussion of these disorders is beyond the scope of this article, emphasis will be placed on two of the more common, potentially treatable, autoimmune causes of sensory neuronopathy: Sjögren’s syndrome and anti-Hu paraneoplastic syndrome. Table 3 shows additional causes of sensory neuronopathy.

Paraneoplastic disorders probably affect less than 1% of all patients with cancer making them extremely rare.135 Although other antibodies and other cancers have been reported with paraneoplastic sensory neuronopathy, anti-Hu antibodies and their high association with small cell lung cancer are the quintessential clinical scenario.136137138139140141142143144 In addition to sensory ataxia, patients may develop concomitant autonomic dysfunction, cerebellar and brainstem involvement, motor neuropathy, and limbic encephalitis.145146 The anti-Hu antibodies, which attack Hu-expressing tumor cells, are thought to trigger a CD8 cytotoxic T cell response.147148149

The sensory neuronopathy sometimes seen in Sjögren’s syndrome is also associated with autonomic dysfunction and at times brainstem dysfunction.78150151 The underlying pathophysiology of Sjögren’s associated sensory neuronopathy is unknown, although T cell mediated infiltration in the DRG has been demonstrated.152

Posterolateral syndrome

Not all sensory ataxic presentations localize to the peripheral nervous system, and disorders affecting the dorsal columns of the spinal cord must also be considered. In contrast to the disorders discussed above, which are mainly autoimmune, the myelopathic disorders that present with sensory ataxia (in addition to spasticity and weakness) often have nutritional or infectious causes (see table 4). Tabes dorsalis, a presentation of parenchymatous neurosyphilis, may selectively affect the dorsal columns and spare the corticospinal tracts.153

Table 4

Myelopathies that present with sensory ataxia*

View this table:

Diagnostic approach

In addition to the clinical examination, the diagnostic evaluation of the sensory polyneuropathies may include a combination of electrodiagnostic studies, testing of autonomic function, laboratory testing, and histopathologic analysis of nerve tissue. Figures 1-3 provide algorithms to guide the diagnostic evaluation of sensory polyneuropathies.

Fig 1
Fig 1

Diagnostic algorithm for neuropathic pain predominant presentations. Abbreviations: ACE=angiotensin converting enzyme; ANA=antinuclear antibody; NCS=nerve conduction studies; QSART=quantitative sudomotor axon reflex test; SPEP=serum protein electrophoresis; SSA=Sjögren’s syndrome A; SSB=Sjögren’s syndrome B; TST=thermoregulatory sweat testing; TTG=tissue transglutaminase.

Fig 2
Fig 2

Diagnostic algorithm for sensory ataxic presentations. Abbreviations: ACE=angiotensin converting enzyme; ANA=antinuclear antibody; CRMP-5=collapsing response mediator protein-5; DADS=distal acquired demyelinating symmetric neuropathy; GBS=Guillain-Barré syndrome; MAG=myelin associated glycoprotein; NCS=nerve conduction studies; SSA=Sjögren’s syndrome A; SSB=Sjögren’s syndrome B; SPEP=serum protein electrophoresis.

Fig 3
Fig 3

Diagnostic algorithm for distal symmetric loss of sensation or neuropathic pain (or both). Abbreviations: NCS=nerve conduction studies. *Based on American Academy of Neurology guidelines.154

Electrodiagnostic studies

Nerve conduction studies

Nerve conduction studies are a sensitive and specific method of assessing disease in the large myelinated nerve fibers and can provide useful diagnostic information regarding the underlying pathophysiology of the neuropathy (fig 4).42154 Most neuromuscular experts advocate for the use of electrodiagnostic studies in distal symmetric polyneuropathy if the diagnosis is known or unknown.155 Several studies have shown that electrodiagnostic studies in this population can often change the diagnosis and management.156157158 Others, however, advocate for its use only in patients with atypical presentations.159 Regardless of this, many clinicians will forgo electrodiagnostic testing in patients who have straightforward distal symmetric polyneuropathy if the underlying cause is known (such as diabetes).

Fig 4
Fig 4

Corneal nerve fiber analysis using corneal confocal microscopy showing (A) healthy control with normal nerve fiber density (arrows) and (B) a patient with diabetic polyneuropathy who has markedly reduced sub-basal nerve fiber density (arrow) and Langerhans cells (arrowheads).

Because electrodiagnostic studies will be normal in disorders that mainly affect small unmyelinated fibers, a normal nerve conduction study does not exclude the presence of small fiber dysfunction. In addition, many disorders with a SFN phenotype may subclinically have involvement of the large myelinated fibers and display abnormalities on electrodiagnostic testing; thus, the presence of large fiber involvement does not exclude small fiber dysfunction.1

In sensory neuronopathies, sensory nerve action potentials (SNAPs) may be absent or display reduced amplitudes with relative preservation of conduction velocities. Abnormalities often do not follow a length dependent pattern and may be widespread. In contrast to most polyneuropathies, the upper extremities may be more prominently affected. Motor studies will classically be normal but subtle abnormalities are often encountered.4445160

The diagnostic criteria for sensory neuronopathies (fig 5), which are based on a large retrospective analysis published in 2009, include at least one absent SNAP or three SNAPs less than 30% of the lower limit of normal in the upper extremities and less than two abnormal motor nerve responses in the lower extremities.46 These criteria were further validated after another large multicenter study was published in 2014.161 A recent case-control study suggests that greater than a 50% difference in amplitude in a side-to-side comparison of two or more pairs of sensory nerves could be used as a rapid screening tool, with sensitivity and specificity greater than 90%.162 Small case series show that blink reflexes may be abnormal in sensory neuronopathies secondary to Sjögren’s syndrome, paraneoplastic disease, and idiopathic sensory neuronopathy, suggesting involvement of the trigeminal ganglion.163164

Fig 5
Fig 5

Diagnostic criteria for sensory neuropathy. CMAP=compound motor action potential; DRG=dorsal root ganglion; EMG=electromyography; LLN=lower limit of normal; MCV=motor nerve conduction velocity; MRI=magnetic resonance imaging; NCS=nerve conduction studies; SNAP=sensory nerve action potential. Adapted, with permission, from Camdessanché and colleagues.46

Evoked potentials

Somatosensory evoked potentials—Somatosensory evoked potentials (SSEPs) evaluate the sensory pathways in both the peripheral and central nervous systems. They are particularly valuable when the proximal portions of the peripheral nerves, which are not studied with routine nerve conduction studies, are affected.165 Bipolar transcutaneous electrical stimulation applied to the skin overlying a selected nerve (often median or tibial) evokes the SSEPs, which are then recorded with standard electroencephalograph scalp disk electrodes. They have an important diagnostic role in CISP, which preferentially affects the nerve roots and proximal nerves and spares the distal sensory nerves.166 Evidence of proximal demyelination is also often apparent in sensory CIDP.123

Laser evoked potentials—Laser evoked potentials (LEPs), which assess the nociceptive pathways both peripherally (Aδ and C fibers) and at the spinothalamic tract centrally, have been called the “most widely agreed upon tool for investigating small fiber damage.”167 A carbon dioxide laser stimulus is applied to the foot and calf. The latency and amplitude of LEPs are measured with scalp electrodes. The pain is perceived as first a prickling sensation (Aδ activation) followed by a dull, burning sensation (C fiber activation). Although LEPs have a high sensitivity (in the 70-80% range) for SFN,167168 there are few laser testing facilities worldwide.168 Given their ease of use, LEPs have been proposed as an alternative to skin biopsy in diabetes associated SFN.167

Quantitative sensory testing

Quantitative sensory testing (QST) can provide evidence of small nerve fiber damage on the basis of the measurement of abnormal sensory thresholds, and because abnormal QST results correlate with abnormalities of IENFD.169170 QST has several limitations, such as its inability to discriminate between central nervous system and peripheral nervous system disease, the need for participant cooperation and attention, and the fact that it may be easily influenced by other factors. Therefore, it should not be used in isolation and needs to be interpreted in the clinical context and in conjunction with other studies.54171172173174

Corneal confocal microscopy

Corneal confocal microscopy is an additional diagnostic tool that enables visualization of the peripheral nerves of the cornea and correlates with IENFD (fig 2). This non-invasive technique uses a combination of corneal nerve fiber length, nerve branch density, and nerve fiber density to evaluate the corneal nerve plexus.175176 It has been shown to detect early small nerve fiber damage in many disorders.175177178179180181182 This technique has advantages over skin biopsy as it is rapid and non-invasive, but it is not yet widely available. There is only a modest correlation with disease stage in any patient and the correlation is of limited utility in clinical practice.183184185 A recent study of nearly 1000 patients with type 1 and type 2 diabetes demonstrated the diagnostic validity of corneal confocal microscopy using a 12.5 mm/mm2 optimal threshold for automated corneal nerve fiber length in type 1 diabetes (73% sensitivity, 69% specificity) and a 12.3 mm/mm2 optimal threshold in type 2 diabetes (69% sensitivity, 63% specificity).176 When considering the entire cohort, a lower threshold for automated corneal nerve fiber length of 8.6 mm/mm2 could rule in diabetic polyneuropathy and an upper threshold of 15.3 mm/mm2 could rule it out (88% specificity, 88% sensitivity). How these studies will be incorporated into clinical practice and their role as a clinical trial outcome measure remain to be determined.176

Autonomic testing

Autonomic testing can help in the diagnosis of SFN, especially when dysautonomia is present.186 Sudomotor function testing as a measure of autonomic function may be assessed through thermoregulatory sweat testing, quantitative sudomotor axon reflex test (QSART), or newer techniques such as electrochemical skin conductance.187 Studies suggest that these autonomic testing modalities provide limited additional diagnostic information when a skin biopsy is abnormal.188

Quantitative sudomotor axon reflex testing

Quantitative sudomotor axon reflex testing is a method of assessing postganglionic sudomotor function through the measurement of local sweat production in predetermined sites (forearm, distal and proximal leg, and foot) in response to iontophoresis of 10% acetylcholine. Abnormal QSART test results have been shown to correlate with decreased IENFD.189 However, a recent moderately sized prospective study found that the addition of QSART to the measurement of IENFD adds little diagnostic value for SFN.188 The limitations of QSART include the technical difficulty of testing, the cost of equipment, and availability.187

Thermoregulatory sweat testing

Thermoregulatory sweat testing measures sweating patterns of the body with the use of an indicator dye in a humidity controlled, heated setting for typically 70 minutes. This technique activates peripheral sudomotor function through central autonomic pathways. Advantages of this test include the topographic analysis of sweat pattern abnormalities and the assessment of both pre-ganglionic and post-ganglionic sudomotor function (when other modalities will be normal in pre-ganglionic lesions). However, this test is technically demanding, requires time commitment on the part of the patient, and is not widely available.187 A recent retrospective study suggests that a novel technique of thermal imaging of forced evaporative cooling corresponds with the results from the standard technique using indicator powder and is more efficient.190

Electrochemical skin conductance

Electrochemical skin conductance has been reported in several small studies as a non-invasive, reliable marker of sweat function and SFN.191192193 Electrical stimulation with low direct voltage current is applied to sudomotor fibers of the palms and soles, which in turn activates sweat glands. However, a recent large systematic review determined that evidence on the use of this technique is limited and of overall poor quality; in addition, it is potentially confounded by technical factors, inconsistent normative values, and funding bias.194

Stimulated skin wrinkling

Stimulated skin wrinkling is the reversible undulation of surface skin that is mediated by post-ganglionic sympathetic fibers. It is tested by immersing glabrous skin (smooth skin without hair, as on the palms or soles of the feet) in water or exposing it to EMLA (eutectic mixture of local anesthetic).195196 It has been shown to correlate with IENFD in patients with sensory polyneuropathy,195197 and it has shown comparable sensitivity to other testing methods for diabetic neuropathy.196


Magnetic resonance imaging

Most patients who present with sensory neuropathy will not benefit from neuroimaging, but in select situations MRI may provide some additional diagnostic benefit. Small case series have demonstrated non-enhancing, longitudinally extensive dorsal column lesions in patients with sensory neuronopathies, indicative of the degeneration of central afferent connections between the DRG and dorsal columns.198 A small case series of patients with CISP suggested that MRI abnormalities such as nerve root enlargement or enhancement may be useful diagnostically in patients with normal nerve conduction study results.125 In patients with posterolateral cord syndrome and sensory dysfunction as a result of dorsal column dysfunction, MRI will often show increased T2 and FLAIR (fluid attenuated inversion recovery) signals at the dorsal columns.

Neuromuscular ultrasound

Neuromuscular ultrasound is an emerging tool that is particularly valuable in immune mediated mixed sensory and motor demyelinating polyneuropathies and in entrapment neuropathies, in which focal nerve enlargement can be detected. In a population of patients with SFN, the sural nerve was found to have a greater cross sectional area compared with healthy controls.199 Currently, most experts do not recommend using neuromuscular ultrasound in patients with pure sensory polyneuropathy, although this field remains ripe for future study.200

Tissue biopsy

Skin biopsy

The European Federation of Neurological Societies/Peripheral Nerve Society Guideline and numerous studies support the use of skin biopsy to assess IEFND and as the gold standard for pathologic diagnosis of SFN (fig 6).201 It is a reproducible and reliable technique with a specificity greater than 90%, sensitivity approaching 80%, and favorable positive and negative predictive values.12202203204 Multiple large cohort studies have been conducted to establish normative values for IENFD at the distal leg because age, ethnicity, and sex are known to produce variations.202203205 A recent longitudinal case-control study showed that rates of IENFD decrease are similar at proximal and distal biopsy sites, regardless of cause, supporting a non-length dependent process.41 Diagnostic criteria for SFN have been proposed to enable patients to be included in clinical trials. Box 1 provides a comparison of the 2008 Devigili criteria and the 2017 Blackmore and Siddiqui criteria (which do not require a skin biopsy).206207 In straightforward cases of SFN, supported by a typical history and examination findings, a skin biopsy is often unnecessary, and further research is needed to elucidate the precise role of skin biopsy in clinical practice.

Fig 6
Fig 6

Skin biopsy (scale bars equate to 50 μM). (A) Healthy control with normal intraepidermal nerve fiber density (white arrow). (B) Patient with diabetic polyneuropathy who has reduced intraepidermal nerve fiber density (white arrowhead) and axonal swellings (black arrow), a common finding in such patients. (C) Patient with diabetic polyneuropathy and severely reduced intraepidermal nerve fiber density (white arrowhead).

Box 1

Proposed diagnostic criteria for small fiber neuropathy

2008 criteria by Devigili and colleagues206

The diagnosis of SFN requires at least two of the following:

  • Clinical signs of small fiber impairment (pinprick and thermal sensory loss, hyperalgesia, or allodynia, or a combination thereof) with a distribution consistent with peripheral neuropathy (length dependent or non-length dependent)

  • Abnormal warm or cooling threshold (or both) at the foot on QST

  • Reduced IENFD at the distal leg

2017 Criteria by Blackmore and Siddiqi207

  • Definite SFN: abnormal neurologic examination (impaired pain or thermal sensation) and any two of QSART, QST, or HRV

  • Probable SFN: abnormal neurologic examination and either QSART, QST, or HRV

  • Possible SFN: abnormal neurologic examination or QSART or QST

  • Abbreviations: HRV=heart rate variability testing; IENFD=intraepidermal nerve fiber density; SFN=small fiber neuropathy; QSART=quantitative sudomotor axon reflex test; QST=quantitative sensory testing.


Nerve biopsy

In general, nerve biopsy is not needed to diagnose patients as having a sensory polyneuropathy, although many of the disorders discussed in this review will have characteristic histopathologic features. In sensory CIDP, nerve biopsy may detect demyelinating features, including hypomyelinated fibers on light microscopy and onion bulb formation, as well as mononuclear cell infiltrates in the interstitial tissue.123 Patients with anti-MAG neuropathies show evidence of demyelination and monoclonal IgM and C3d deposits on myelin sheaths.208 Ultrastructural studies show widening of the myelin lamella due to M-protein and activated complement proteins, which colocalize with MAG in these areas.208209210211212 Although a diagnosis of sensory neuronopathy is considered “definite” only if there is pathologic evidence of DRG degeneration, DRG biopsy is discouraged because of the associated morbidity.45161

Current disease specific treatments

Apart from SFN associated with diabetes and prediabetes, the sensory polyneuropathies discussed are relatively rare, and no universally accepted disease specific treatments exist. Many of the disease specific treatments discussed below are based on expert opinion, retrospective studies, and small prospective studies, rather than large randomized placebo controlled trials. Some treatments discussed are emerging and in various stages of study.

Small fiber neuropathies


Evidence to support the optimal treatment regimen for SFN associated with sarcoidosis is limited. In a retrospective review of 115 patients, the SFN treatment response rates were 76%, 67%, and 71% for treatment with intravenous immunoglobulins, anti-TNF-α, and combination therapy with both, respectively.75 By contrast, in the same trial patients treated with methotrexate or corticosteroids showed no improvement or even worsening of symptoms.

Transthyretin familial amyloidosis polyneuropathy

The US Food and Drug Administration and the European Commission have recently approved patisiran and inotersen as treatments for TTR-FAP. Several other drugs, such as diflunisal and tafadimis, have shown promising results in large randomized placebo controlled clinical trials. Liver transplantation has traditionally been the standard treatment despite continued deposition of wild-type transthyretin.213 Patisiran is an RNA interference therapeutic agent that inhibits hepatic synthesis of transthyretin.214 In a double blind placebo controlled phase III trial, 225 patients were randomized to either intravenous patisiran (0.3 mg/kg/body weight) or placebo every three weeks. Patients receiving patisiran had a significant improvement in the Modified Neuropathy Impairment Score +7 (mNIS+7) (P<0.001), on the Norfolk Quality of Life-Diabetic Neuropathy (Norfolk QOL-DN) questionnaire (P<0.001), and gait speed (P<0.001). In addition, a large phase III randomized double blind placebo controlled trial of inotersen, an antisense oligonucleotide that inhibits the hepatic production of transthyretin, has recently been published.215 One hundred and seventy two patients(112 in the inotersen group and 60 in the placebo group) were given weekly subcutaneous injections for 66 weeks. As in the patisiran trial, the treatment arm also significantly improved on the mNIS+7 and the Norfolk QOL-DN scores (both P<0.001). However, inotersen was associated with thrombocytopenia and glomerulonephritis in some patients.

The transthyretin tetramer stabilizers include diflunisal and tafadimis. Diflunisal, a non-steroidal anti-inflammatory drug, strongly inhibits TTR amyloid fibril formation. A large international double blind placebo controlled trial of 130 patients found that diflunisal slowed the progression of patients with and without the TTR-Val30Met and non-Val30Met mutations.216 This orphan drug is widely available and inexpensive. Another randomized double blind placebo controlled trial studied tafamidis in patients with early stage TTR.217 Although the coprimary endpoints of slowed progression on the Neuropathy Impairment Score-Lower Limbs (NIS-LL) (as determined by NIS-LL response, “responders” had an increase in NIS-LL at 18 months of <2 points) and Norfolk QOL-DN scores were not reached, a statistically significant 52% reduction in the worsening of neurologic function (as determined by change in NIS-LL from baseline to 18 months) was seen in this intention to treat population (P=0.027). This drug is approved in Europe, South America, and Japan, but not in the US.89

Sensory ataxic neuropathies

Miller-Fisher syndrome

A retrospective study and expert opinion indicate that intravenous immunoglobulin probably reduces the time to recovery and prevents the progression of symptoms.218219 However, the use of such an expensive treatment in a condition with a favorable prognosis is controversial.33218 An evidence based guideline report of the therapeutics and technology assessment subcommittee of the American Academy of Neurology stated that there was insufficient evidence to support or refute the use of intravenous immunoglobulin in this condition,220 although patients with considerable overlap with GBS should be offered treatment.

Chronic ataxic neuropathy with disialosyl antibodies (CANDA)

Data to guide treatment in these patients are limited.221222223224 In case series, intravenous immunoglobulins have been used with some success,119 whereas rituximab was the most effective treatment in one small cohort of patients, halting disease in eight of nine patients.117

Sensory chronic inflammatory demyelinating polyneuropathy (CIDP)

It is extremely important to recognize this disease because 90% of patients responded to immunotherapy in one series.123 No prospective randomized placebo controlled trials have studied immunosuppressant or immunomodulatory therapy in the sensory variant of CIDP specifically. In one retrospective series of 15 patients with CISP, all patients responded to intravenous immunoglobulins or intravenous methylprednisolone.125

Distal acquired demyelinating sensory neuropathy (DADS)

Many treatments have been tried and abandoned in MAG neuropathies including corticosteroids, intravenous immunoglobulins, and plasma exchange.212 Although cytotoxic agents such as fludarabine, cyclophosphamide, and chlorambucil may be beneficial, their toxicities limit longterm use.212 Rituximab, a monoclonal antibody that targets CD20 (a B cell surface antigen) and depletes circulating B cells, has been used with success in 30-50% of patients in uncontrolled trials.128225 The primary endpoints in two placebo controlled randomized trials of rituximab failed to reach statistical significance, although secondary endpoints such as time-to-walk scales significantly improved.225226227 Patients with motor deficits and subacute progression may respond more favorably to rituximab if the drug is started soon after the onset of symptoms.228229 In 2010, the European Federation of Neurological Societies and Peripheral Nerve Society published a guideline on the management of paraproteinemic demyelinating neuropathies. This guideline highlighted the lack of proven efficacy for any therapy in anti-MAG neuropathy but emphasized that some patients may respond to treatment.230 Two patients have been treated with obinutuzumab, a first generation glycoengineered type-I, anti-CD20 mediated, B cell depleting monoclonal antibody.231 No improvement or worsening in the patients’ neuropathic symptoms was seen after 12 months of treatment.

Sensory neuronopathies

Given the rarity of these diseases, little is known about the best approach to treatment, although a treatment window probably exists. A case series of serial nerve conduction studies in patients with sensory neuronopathy suggests that sensory abnormalities plateau after 7-10 months from symptom onset. On the basis of the rate of decline of sensory response amplitudes, treatment should be started within the first eight months if possible.232 Beyond this window, the inflammatory reaction probably dampens and treatment becomes unsuccessful. For the patients with paraneoplastic disease, detection and treatment of the underlying cancer is obligatory. For both paraneoplastic and Sjögren’s associated sensory neuronopathies, immunosuppressant and immunomodulatory treatment should be provided. Intravenous immunoglobulin, plasma exchange, corticosteroids, rituximab, cyclophosphamide, infliximab, and azathioprine have all been used in uncontrolled studies of Sjögren’s associated sensory neuronopathy.233234235236237238 Corticosteroids,239240 intravenous immunoglobulins,241242 plasma exchange,243 rituximab,244 and sirolimus245 have been used in patients with paraneoplastic sensory neuronopathy.

Emerging disease specific treatments

Small fiber neuropathies

Diabetes and prediabetes

In early clinical trials, physical exercise has shown promise as a treatment of SFN associated with glucose dysregulation. In prospective randomized trials, exercise results in increased IENFD in patients with diabetes but no neuropathy.246247 A small prospective pilot study in diabetic neuropathy also found that pain responded to exercise.248 A large prospective randomized study of patients with type 2 diabetes associated peripheral neuropathy (the Activity for Diabetic Polyneuropathy or “ADAPT” study), which is investigating the effect of supervised exercise versus standard care counseling on polyneuropathy, as measured by IENFD and change in quality of life, is currently underway (Clinical trials identifier NCT02341261).


ARA 290 (Cibinetide), an erythropoietin derivative that activates the innate repair receptor and initiates anti-inflammation, cytoprotection, and healing, has been well tolerated and showed benefit in treating sarcoidosis associated neuropathic pain in two phase II clinical trials.18249

Sjögren’s syndrome

Like all polyneuropathies that are associated with Sjögren’s syndrome, studies of the disease specific treatment of Sjogren’s associated SFN are sparse. A small uncontrolled trial of intravenous immunoglobulins in Sjögren’s syndrome found a reduction in neuropathic pain in SFN,250 but in other series the response to corticosteroids has been poor.150251252 A small prospective, phase III clinical trial testing the benefit of intravenous immunoglobulins in patients with painful large fiber sensory polyneuropathy will soon be enrolling and could potentially inform treatment choices in patients with SFN that is associated with Sjögren’s syndrome (Clinical trials identifier NCT03700138).

Sodium channelopathies

The discovery of the SCN9A, SCN10A, and SCN11A genetic mutations has opened the door to potential therapeutic options for painful SFN.88253254 Lacosamide, a blocker of Nav1.3, Nav1.7, and Nav1.8 that stabilizes channels in the slow inactivation state, has been studied in SCN9A associated SFN, although the results have not yet been published255 (Clinical trials identifier NCT01911975).

Current management of neuropathic pain

Neuropathic pain and positive sensory disturbances contribute greatly to the morbidity associated with sensory polyneuropathy. Most studies have focused on the treatment of painful neuropathy secondary to diabetes or chemotherapy induced painful neuropathy. A large meta-analysis published in 2015 updated recommendations on the pharmacologic treatment of neuropathic pain.256 This review found moderate to high quality of evidence for the use of serotonin-norepinephrine reuptake inhibitors (SNRIs), pregabalin and gabapentin, tricyclic antidepressants (TCAs), opioids, botulinum toxin, and capsaicin. SNRIs, TCAs, gabapentin, and pregabalin were given a strong recommendation and proposed as first line agents, whereas topical capsaicin or lidocaine and tramadol were given a weaker recommendation and proposed as second line. Strong opioids and botulinum toxin A were recommended as third line.

A recent large retrospective systematic review of 106 randomized controlled trials examined the effect of various drugs for diabetic neuropathy on pain and quality of life. Anticonvulsants including pregabalin and oxcarbazepine; SNRIs including duloxetine and venlafaxine; TCAs; atypical opioids including tramadol and tapentadol; and botulinum toxin A were determined to be more effective than placebo. The strength of evidence was considered moderate for SNRIs and low for the other listed agents. The review concluded that other commonly used agents including gabapentin, topical capsaicin, typical opioids, dextromethorphan, and mexiletine were no more effective than placebo.257

A large multicenter double blind parallel group study of diabetic neuropathic pain studied whether patients who did not respond to standard dose monotherapy with duloxetine (60 mg/day) or pregabalin (300 mg/day) would respond to high dose duloxetine (120 mg/day), high dose pregabalin (600 mg/day), or a combination of both (duloxetine 60 mg/day and pregabalin 300 mg/day). Eight hundred and four patients were evaluated for initial monotherapy, and the 339 who were considered non-responders were treated with high dose monotherapy or combination therapy. The primary outcome measure was the Brief Pain Inventory Modified Short Form (BPI-MSF) 24 hour average pain change after starting high dose monotherapy or combination therapy. No statistically significant differences in the BPI-MSF average pain score were seen between the combination and high dose monotherapy groups (P=0.370). When the initial standard monontherapy doses were compared, 60 mg/day of duloxetine was superior to 300 mg/day of pregabalin (P<0.01).258

A prospective interventional study conducted in an Indian diabetic clinic enrolled 100 patients who had never been treated with drugs for neuropathic pain to receive either pregabalin (50 patients) or duloxetine (50 patients). Comparative efficacy was determined by the Neuropathic Pain Scale (NPS) and the Neuro-QOL, a quality of life instrument. On the basis of NPS and Neuro-QOL scores, the efficacy of duloxetine was 1.27 and 1.44 times that of pregabalin, respectively. Cost effectiveness was calculated using cost consequence analysis, the average cost effectiveness ratio and the incremental cost effectiveness ratio. This analysis demonstrated that duloxetine, while slightly more expensive, demonstrated a significant improvement in quality of life.259

The use of opioids for the management of chronic neuropathic pain is generally discouraged. Although these drugs are efficacious in the short term, evidence to support their longer term use is weaker, and serious safety concerns exist.260 A recent meta-analysis of 96 randomized placebo controlled trials investigating the efficacy of opioids for chronic non-cancer pain found that opioid use was associated with significant but small improvements in pain and physical functioning. When opioids were compared with non-opioid alternatives the benefit for pain and physical functioning seemed to be similar.261

Neuropathic pain management guidelines

Both the European Federation of Neurological Societies (EFNS) and American Academy of Neurology (AAN) (updated 2010 and 2011, respectively) have published guidelines on the pharmacologic management of painful diabetic peripheral neuropathy (table 5).262263 Both guidelines support the use of TCAs, pregabalin, gabapentin, various opioids, SNRIs, and topical lidocaine for the treatment of neuropathic pain. The AAN guidelines also recommend the use of topical capsaicin and valproate with level B evidence and only pregabalin was supported by level A evidence. Each of these guidelines recommend against the use of oxcarbamazepine, lamotrigine, lacosamide, clonidine, and mexiletine in the symptomatic management of painful neuropathy.

Table 5

Drugs used to treat neuropathic pain*

View this table:

Emerging treatments for neuropathic pain

Chemodenervation with botulinum toxin A is thought to inhibit the release of peripheral neurotransmitters such as acetylcholine and nociceptive peptides (substance P, glutamate, calcitonin gene related peptide) from sensory nerves.264 In addition, botulinum toxin inhibits vanilloid receptor TRPV1 expression on the surface of peripheral nociceptors. It has been supported by multiple studies as described in a recent review for the treatment of neuropathic pain.265

A 2015 meta-analysis evaluating the use of several treatments for neuropathic pain gave botulinum toxin A a weak recommendation.256 In the same year, a meta-analysis looking at chemodenervation in diabetic peripheral polyneuropathy supported its use as a result of finding clinically significant improvements in pain scores.264 Most recently, in 2017, a large systematic review found that botulinum toxin was more effective than placebo, although the strength of evidence was low.257

There is some evidence from small studies to support the use of acupuncture as a non-pharmacologic treatment for neuropathic pain. A randomized placebo controlled partially blinded trial in Germany is currently looking at the effect of needle acupuncture, laser acupuncture, and placebo laser acupuncture on electrophysiologic parameters, neurologic deficits, and symptoms.266


The sensory polyneuropathies are heterogeneous conditions with distinct clinical phenotypes defined by the type of nerve fibers involved and the time course. Although many of the small fiber, pain predominant and large fiber, ataxia predominant neuropathies discussed are relatively uncommon, those reviewed have the shared feature of being potentially treatable and even reversible. The recognition of these distinctive presentations is of utmost importance to enable treatment to be started before permanent nerve damage occurs.

Questions for future research

  • Because impaired glucose tolerance and type 2 diabetes are highly associated with neuropathy, further study into the best way to treat hyperglycemia (and other risk factors such as central obesity and hypertriglyceridemia) is needed. This is extremely important given the morbidity and disability associated with diabetic neuropathy, the most common cause of neuropathy worldwide.

  • In transthyretin familial amyloidosis with polyneuropathy further studies are needed to determine the effect of inotersen and patisiran on the cardiomyopathy. In addition, we need to find a way to identify this rare subset of patients early in the disease course to avoid delays in start of treatment.

  • Advances in understanding the role of antiganglioside antibodies in the sensory neuropathies have recently been made. How will an even greater understanding of the underlying pathomechanisms of these ganglioside antibodies translate to more targeted therapies?

How patients were involved in the creation of this article

Two patients provided their experience of having sensory neuropathy in their own words. They were given the option to review the manuscript but declined.


Thanks to A Gordon Smith, Peter Hauer, and Stormy Foster-Palmer for contributing the figures.


  • Series explanation: State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors.

  • Contributors: KGG and KP performed the literature review and prepared the initial draft of the manuscript. Both authors were involved in the conception, drafting, and editing of the manuscript. KGG is guarantor.

  • Competing interests: The authors have read and understood the BMJ policy on declaration of interests and have no competing interests.

  • Provenance and peer review: Commissioned; externally peer reviewed.


View Abstract