Clinical manifestations

Presentation and course

Sensory neuronopathies share a general clinical and electrophysiological pattern which is typically that of a purely sensory neuropathy with a nonlength dependent distribution of sensory loss (06; 31). The four limbs are usually affected mostly in their distal part (feet and hands). The disorder may in addition involve the face and the trunk. The loss of large sensory neurons results in ataxia that usually predominates in the lower limbs, resulting in ataxic gate and Romberg sign. In some cases, the disorder is so severe that the patient is unable to stand up and walk. Deep or superficial sensory loss also affects the upper limbs, which may result in severe disability, with some patients being unable to perform any purposeful movement with their hands. Pseudo athetoid movements of fingers or hands are a manifestation of deep sensation loss. Damage to small neurons induces a loss of thermal and pain sensation. Neuropathic pain is frequent in this case, and trophic skin lesion may appear, particularly with genetic sensory neuronopathies.

The initial distribution of sensory disturbances and their extension with time are strongly dependent on the underlying mechanism of the disease, which helps to orientate the etiological diagnosis. Thus, a symmetrical distribution starting in the lower limbs and a prominence of deep sensation loss are common with genetic or toxic sensory neuronopathies but may also be encountered in some idiopathic cases, particularly in the elderly. In opposition, asymmetry or a multifocal distribution sometimes mimicking mononeuritis multiplex and early involvement of the hands or face characterize inflammatory sensory neuronopathies including paraneoplastic sensory neuronopathies and sensory neuronopathies associated with autoimmune diseases. But with time, the distribution usually becomes more symmetrical and diffuse. An acute or subacute evolution characterizes chemotherapy-induced, infectious, or paraneoplastic sensory neuronopathies whereas a progressive course is the hallmark of genetic diseases but also occurs in dysimmune or idiopathic sensory neuronopathies.

Prognosis and complications

Prognosis is quite variable. Some patients may have a restricted and limited form, whereas others may develop a very disabling disorder and become chair-bound. Acute form, sometimes classified as sensory Guillain-Barré syndrome, may not recover. Association with a severe autonomic neuropathy may have a bad prognosis because patients develop severe orthostatic hypotension, cardiac arrest, pseudo-digestive obstruction (typically paraneoplastic sensory neuronopathies), or sudomotor crisis with intolerance to heath. Small fiber involvement may result in chronic skin ulcerations and indolent burns.

Clinical vignette

A 53-year-old woman with chronic alcoholic consumption and smoking complained of two years of eye dryness. In March, paresthesia and dysesthesia appeared within a few days in the first three fingers of her right hand and rapidly reached all the others. A few weeks later symptoms extended to the left hand. She experienced difficulties with buttons, writing, and manipulating small objects due to sensory loss and pain elicited by contact. In April, similar symptoms appeared in the feet and gait became unsteady. The patient was referred in June. Although her gate was slightly ataxic and symptoms persisted in the hand, the patient was totally independent. Tendon reflexes were diffusely abolished. Vibration sense was abolished and pain sensation reduced in the four extremities. Electroneuromyography showed a severe reduction of all the tested sensory action potentials in the four limbs whereas motor conduction velocities were normal. Such a pattern, which fulfills the diagnostic criteria of possible sensory neuronopathies (score = 12.7) (see infra) ruled out an alcoholic length-dependent polyneuropathy. A diagnosis of sensory neuronopathy was established and a biological workup in search of the underlying etiology was performed and revealed to be negative: there was no sicca syndrome, no monoclonal gammopathy. Sjögren syndrome associated anti-SS-A and SS-B antibodies were negative as were the search for paraneoplastic antibodies (Hu, Ma2, CRMP5, amphiphysin, Yo, Ri antibodies), antibodies associated with hepatitis, celiac disease, and antiganglioside antibodies. Viral serology was negative. However, CSF examination showed normal protein concentration with an oligoclonal pattern of IgGs and 10 lymphocytes/mm3 (N < 1). This joined with the subacute onset of the disorder and the presence of pain led to suspect a seronegative paraneoplastic sensory neuronopathy. FDG-PET scanner was performed and revealed an abnormal uptake in the left breast. Biopsy disclosed an adenocarcinoma. The patient was treated for her cancer. The neuropathy slightly improved and then remained stable. A final diagnosis of paraneoplastic sensory neuronopathy was retained.

Biological basis

Etiology and pathogenesis

Paraneoplastic sensory neuronopathies: subacute paraneoplastic sensory neuronopathy. Sensory neuronopathy is the most frequent paraneoplastic neurologic syndrome, representing 24% of cases in the series of about 1000 paraneoplastic patients of the PNS Euronetwork group (20). It is thought to be a T cell mediated disorder targeting antigens expressed by the tumor, which are also expressed by neurons in dorsal root ganglia (14). The most frequent of these onconeural antigens is HuD (23). The onset is subacute or rapidly progressive but indolent and protracted courses over several months may occur (22). Sensory loss is frequently multifocal or asymmetrical, and as a rule, involves the upper limbs in a nonlength dependent distribution. The face, chest, or trunk can also be concerned. Although large and small sensory neurons are simultaneously affected, in some cases, lesions predominate on one type of neurons resulting in a mostly ataxic sensory neuronopathy or a mostly painful small fiber neuropathy. The autonomous nervous system is involved in 20% to 24% of patients and manifests as gastrointestinal dysmotility, orthostatic hypotension, arrhythmia, or urinary dysfunction (23).

On the electrophysiological point of view, beside a severe alteration of sensory action potentials, mild modifications of motor conduction velocities and neurogenic needle recordings are not infrequent (07; 54). These changes result from subclinical involvement of lower motor neurons in the spinal cord and the extension of the inflammatory reaction into the roots and peripheral nerves. The CSF usually shows elevated protein concentration, pleocytosis, and oligoclonal bands. In about 90% of cases sensory neuronopathy precedes by several months the appearance of the underlying tumor or its relapse. In 70% to 80% of cases, it is a small cell lung cancer. Most patients have Hu antibodies. Amphiphysin or CV2/CRMP5 antibodies occur in 5% to 10% of cases with or without Hu antibodies and in 10% to 16% of patients onconeural antibodies are not found. These patients usually have another type of tumor including breast cancer or Hodgkin disease (20).

The management of paraneoplastic sensory neuronopathies is difficult. Immunomodulatory treatments including prednisolone, intravenous immunoglobulins, plasma exchanges, cyclophosphamide, tacrolimus, and rituximab alone or in combination have been tested in open and uncontrolled small series. Their efficacy is not clearly proven although some patients improve (03). Tumor treatment is probably the best way of stabilizing the disease so that all the effort should be made to obtain an early diagnosis of the cancer (23). Among possible investigations, FDG-PET scanner is certainly the first-line exam. In a patient with onconeural antibodies, if the initial workup is negative it should be repeated after three months and then every six months for four years (70).

Toxic sensory neuronopathy. High doses of pyridoxine have been demonstrated to induce sensory neurons necrosis in rats (62). In humans, chronic consumption of pyridoxine induces an irreversible ataxic neuropathy. The effect is probably dose-dependent and the toxic dose may vary between 200 and 6000 mg/day.

The most frequent cause of toxic sensory neuronopathy depends on the use of platinum salt for the treatment of cancer (30). Cisplatin and oxaliplatin are more toxic than carboplatin (38). The cumulative toxic dose is 300 mg/m2 for cisplatin and oxaliplatin and 400 mg/m2 for carboplatin. The onset is usually subacute, with severe paresthesia and ataxia in the four limbs. Oxaliplatin also induces transient mouth and throat sensory symptoms aggravated by exposure to cold in the hours following infusion (38). A so-called coasting phenomenon due to a delayed release of the toxic drug accumulated in tissues is responsible for the development or worsening of the neuropathy up to several weeks after treatment interruption so that an arrest of treatment should be considered as soon as progressing sensory symptoms appear (02). In animals, platinum salt induces multiple nuclear and mitochondrial DNA damages that may eventually result in sensory neurons apoptosis (43; 68). This explains that the clinical recovery is frequently incomplete. In patients with paraneoplastic sensory neuronopathy, the use of platinum salt for the treatment of cancer may be problematic, and treatment with the less toxic compound should probably be preferred.

Infectious sensory neuronopathy. This is probably a rare condition. The most numerous cases, including pathological demonstration of inflammatory changes in the dorsal root ganglia, have been published in the setting of HIV infection before the era of antiviral treatments. The onset is acute or chronic although the viral gp120 protein can bind the neurons. The mechanism is more probably an autoimmune disorder triggered by the virus present in endoneurial macrophages (13). Occasional cases of an acute nonlength dependent pure sensory neuropathy, thus probably true sensory neuronopathy, have been reported with Epstein-Barr virus (59), human T-lymphotropic virus (40), herpes zoster virus, or Zika virus infection (41).

Sensory neuronopathies with systemic autoimmune diseases. Sensory neuronopathies have mostly been reported in association with Sjögren syndrome (24; 47) and more rarely with autoimmune hepatitis (46; ), celiac disease (17; 26), lupus, or other unclassified connective diseases. The best studied cases are those with Sjögren syndrome for which several series have been published, some with pathological demonstration of specific inflammatory dorsal root ganglia lesions (24; 47). The onset is usually chronic and symmetrical, but some patients may follow a more rapid course. In most cases, ataxia and loss of proprioception are predominant but some patients, may have pure or predominant involvement of small neurons with pain (47; 21; 27). In both cases the distribution is similar, with constant involvement of the four limbs and in 30% to 60% of cases of the face or trunk. Autonomic neuropathy including Adie or Argyll Robertson pupils, orthostatic hypotension, and anhidrosis occur in about 70% of patients (47). Although it has not been demonstrated that the painful form is associated with dorsal root ganglia lesions, this is highly probable. Changes in dorsal root ganglia are relatively similar to that observed in paraneoplastic sensory neuronopathy with an infiltration of T cells in contact of degenerating sensory neurons (24; 47). Sensory neuronopathy frequently precedes by months of years the diagnosis of Sjögren syndrome so that investigation must be regularly renewed in search of a sicca syndrome or SS-A or SS-B antibody positivity. As with paraneoplastic sensory neuronopathy, different immunomodulatory treatments have been used in retrospective and uncontrolled series. Some patients are said to improve with intravenous immunoglobulins or steroids (47; 57). Immunosuppressants should probably be used with some caution in patients with Sjögren syndrome because lymphoma occurs with a higher frequency in this disorder.

Sensory neuronopathies and antidisialosyl ganglioside antibodies. Polyclonal IgM antibodies directed toward disialosyl ganglioside including GD1b, GT1a, and GD3 occur more frequently in chronic ataxic sensory neuropathies (64) whereas antidisialosyl ganglioside monoclonal IgM recognizing in addition to GQ1b are associated with CANOMAD (chronic ataxic neuropathy, ophthalmoplegia, M-protein, cold agglutination, and antidisialosyl antibodies) (73). These disorders are probably heterogeneous because the electroneuromyography may alternatively show features consistent with a primary demyelinating process or a sensory neuronopathy. In addition, autopsy studies in two cases with antidisialosyl antibodies found predominant demyelination in the dorsal roots (53), whereas the other showed involvement of neurons in the dorsal root ganglia (44). Rabbit immunization with GD1b has been reported to induce dorsal root ganglia lesions, but this could not be reproduced (32). Thus, the exact underlying process in these disorders remains unclear.

Idiopathic sensory neuronopathy. The majority of sporadic sensory neuronopathies remain idiopathic after an extensive workup (05). These cases may present with an acute (74; 28), subacute (61), or slowly progressive course (12). Some patients may have subtle findings of an underlying unspecific associated immune context such as low levels of antibodies associated with systemic autoimmune disease or oligoclonal bands in the CSF (60). These neuropathies may also occur with autonomic neuropathy. Before considering a sensory neuronopathy as idiopathic it should be kept in mind that genetic disease such as Friedreich ataxia (56), or CANVAS (cerebellar ataxia, neuropathy, vestibular areflexia syndrome) may present in adult patients without family history (67). Similarly, the sicca syndrome may appear several years after the onset of sensory neuronopathy, and the criteria of Sjögren syndrome may be reached very late in the evolution of the neuropathy (47). CANVAS is now recognized as a frequent cause of sporadic sensory neuronopathy of adult onset. In some series, it may account for 30% of apparently idiopathic cases. Preserved tendon reflexes and a long preceding history of chronic cough are suggestive of CANVAS. Things are complicated by the fact that patients with CANVAS may also develop Sjögren syndrome (11).

A subgroup of idiopathic sensory neuronopathy is probably immune-mediated, as shown by a few pathological studies of dorsal root ganglia. Changes are very similar to that observed in paraneoplastic sensory neuronopathy or sensory neuronopathy with Sjögren syndrome with degeneration of sensory neurons in presence of infiltrating lymphocytes, which when tested are cytotoxic T-cells (10; 28). These cases correspond to acute, subacute, or chronic sensory neuronopathy with or without autonomic neuropathy, and they may account for about 30% of apparently idiopathic cases (10; 60). The difficulty is that only a biopsy of a dorsal root ganglion may theoretically confirm the diagnosis, which of course cannot be recommended. Thus, there is a need for biomarkers, and two antibodies have been proposed as such biomarkers.

Antibodies toward the intracellular domain of the fibroblast growth factor receptor 3 (FGFR3) have been reported in a series of patients with sensory neuropathy (01; 69). Patients with this antibody have a nonlength dependent sensory neuropathy in 89% of cases and met the criteria for sensory neuronopathy in 64% of cases and small fiber neuropathy in 17% of cases (69). The antibodies were the only marker of disimmunity in 66% of patients, and the course was mostly chronic. Other studies have linked anti-FGFR3 antibodies to a broader range of neuropathies (29; 52; 71). The methodology employed for antibody detection is undoubtedly critical for ensuring result specificity (49). Some reports have suggested that patients with anti-FGFR3 antibodies may improve with intravenous immunoglobulins (29). However, the study of Tholance and colleagues did not confirm this finding, as there was no superiority of steroids, intravenous immunoglobulins, or immunosuppressants over each other in patients who received immunomodulatory treatments (69).

Antibodies against argonaut (AGO) proteins 1 and 2, initially described in patients with lupus or Sjögren syndrome, have been identified in inflammatory neurologic disorders, Sensory neuronopathy is the most frequent among them (16; 50). Compared to patients with sensory neuronopathy and no anti-AGO antibodies, those with anti-AGO antibodies exhibited a more severe disease and a higher likelihood of developing Sjögren syndrome. Interestingly, patients with anti-AGO antibodies showed a more favorable response to intravenous immunoglobulin, and the presence of these antibodies was predictive of treatment response. These findings, based on a limited number of patients, require confirmation in larger studies.

Both anti-FGFR3 and anti-AGO antibodies are not specific to sensory neuronopathy, but they are probably helpful to identify an otherwise unapparent underlying disimmune background in patients with apparently idiopathic sensory neuronopathy.

Epidemiology

Most sensory neuronopathies are acquired. However, there is no epidemiological study providing their exact frequency and repartition among etiologies. In a French multicenter retrospective study of 136 patients, paraneoplastic sensory neuronopathy represented 11%, sensory neuronopathy associated with a dysimmune disease 19.8%, toxic sensory neuronopathy 12.5%, and idiopathic sensory neuronopathy 39.7% of cases (05). The recent identification of CANVAS certainly increases the proportion of patients with an apparently idiopathic sensory neuronopathy who have a genetic disease.

Prevention

Early diagnosis and treatment of associated conditions may possibly stabilize the disease.

Differential diagnosis

Differential diagnosis mostly includes other ataxic neuropathies and nonacquired sensory neuronopathies. CIDP or anti-MAG monoclonal IgM neuropathies are not difficult to differentiate when the electroneuromyography is clearly demyelinating. Things are more difficult with the proximal form of ataxic CIDP in which demyelination is almost limited to the plexus or roots (66). In this case, motor conduction velocities may be normal but sensory action potentials are not severely altered and somatosensory evoked potentials show delayed potential at the level of plexus or spinal cord entry. Plexus and root MRI is probably the best tool to diagnose these forms by showing enlarged plexus or nerve roots with gadolinium enhancement. However, it may sometimes be very difficult to differentiate sensory neuronopathies from pure sensory CIDP. In this case nerve biopsy may be recommended. B12 deficiency may also mimic sensory neuronopathies all the more so because T2-weighted MRI usually shows hyperintensities in the posterior column of the cervical spinal cord. Sensory nerve vasculitis may also be difficult to differentiate, as acquired inflammatory sensory neuronopathy can have an initial multifocal distribution. The differential diagnosis may even be complicated by the fact that nerve biopsy may show vasculitis in dysimmune sensory neuronopathy.

It may sometimes be difficult to distinguish acquired from hereditary sensory neuronopathy. Sensory neurons are involved in a host of genetic diseases but usually as a component of a complex disorder that involves different neural or nonneural structures. Some of them may present as ataxic sensory neuronopathy over several months or years before extension of symptoms to other systems and therefore may be misdiagnosed as acquired sensory neuronopathy (Table 1), particularly with an adult onset and a presentation of sporadic case. This particularly occurs with Friedreich ataxia (55; 56), SANDO syndrome due to POLG mutation (63; 36), and CANVAS (67; 58; 11). It may be difficult to differentiate clinically cerebellar ataxia from proprioceptive ataxia when the two simultaneously occur as in Friedreich ataxia and CANVAS. CANVAS may present as recessive or sporadic cases with late onset sensory neuronopathy. Vestibular areflexia is inconstant and chronic cough present in about one third of cases may precede the onset of the neuropathy by several years (11). Very rare cases of sensory neuronopathy have been reported in the setting of hereditary amyloid neuropathy due to transthyretin gene mutations (39).

Table 1. Hereditary/Genetic Sensory Neuronopathies That May Present at Least For a Time as Sporadic Cases of Adult Onset Sensory Neuropathy

Diagnostic workup

Paraclinical investigations for the diagnosis of sensory neuronopathies. Electroneuromyography typically shows a severe reduction or absence of sensory action potentials in the four limbs with normal or almost normal motor conduction velocities (06). However, subtle motor abnormalities may occur with most of the causes of sensory neuronopathy as minor to mild reduction of motor action potentials and motor conduction velocities or presence of mild neurogenic changes on needle examination (07; 08). These motor changes should be interpreted after taking into account the clinical pattern of the neuropathy and the severity of sensory action potential abnormalities to not miss the diagnosis of sensory neuronopathy .

Sensory evoked potentials have been proposed to investigate the dorsal root ganglia because they can explore the proximal peripheral process of large sensory neurons from plexus to spinal cord entry and their spinal cord central process (34). However, they need a specific recording methodology (51) and the persistence of a sufficient number of sensory neurons to be elicited. In addition, it may be difficult to differentiate a demyelinating disorder of the proximal part of peripheral nerves from a disorder affecting the dorsal root ganglia.

Laser-evoked potentials (37) and skin biopsy (35; 21) can be used to demonstrate a nonlength dependent distribution of small fiber loss. But if nonlength dependent small fiber neuropathies have been attributed to dorsal root ganglia involvement, this has not been pathologically demonstrated yet.

Spinal cord MRI (09) may show the degeneration of the central process of large sensory neurons in the dorsal column appearing as a T2-weighted signal hyperintensity at the level of the cervical spinal cord (33; 48; 19). This may be associated to some degree with spinal cord atrophy (19). However, this is inconstant and not specific to sensory neuronopathy because, for example, B12 deficiency induces T2-weighted hypersignal.

Nerve biopsy is not useful for the positive diagnosis of sensory neuronopathy (12). When performed it shows fiber loss and ongoing axonal degeneration whose severity depends on the course of the neuropathy. It is more interesting for the differential diagnosis with forms of purely sensory chronic inflammatory demyelinating polyneuropathy or vasculitis. Only dorsal root ganglia biopsy may be able to demonstrate the degeneration of sensory neurons with a secondary proliferation of satellite glial cells surrounding neurons resulting in the formation of Nageotte nodules. However, dorsal root ganglia biopsy cannot be recommended as a routine procedure.

The diagnostic strategy of sensory neuronopathies.

Differentiating sensory neuronopathies from other sensory neuropathies. Asbury once proposed that a nonlength dependent sensory neuropathy with pure or almost pure and severe sensory action potential alteration is a sensory neuronopathy but these criteria were never tested (06). We published another set of criteria (Table 2) (08) which tested on a large external population of patients with sensory neuropathy originating from neuromuscular diseases centers, who proved to have 90% sensitivity and 85% specificity against the expert diagnosis (05). These criteria only rely on clinical and electroneuromyography data that are easy to obtain.

Table 2. Diagnostic Criteria of Sensory Neuronopathy

Searching for the underlying cause of the sensory neuronopathy. The second step is the etiological diagnosis once the diagnosis of sensory neuronopathy is established. A study addressed the question of the etiological diagnosis of sensory neuronopathy and proposed a strategy that was checked against an internal and external population using multiple correspondence analysis with 95% sensitivity (see Figure 1 below). In a first step a patient with sensory neuronopathy according to the criteria should be investigated with a least a determination of B12, SS-A, SS-B, hepatitis associated antibodies, FGFR3 and AGO antibodies, onconeural antibodies and monoclonal gammopathy. A positivity of one of these tests orientates the etiological diagnosis. If this workup is negative, one has to take into account the evolution of the sensory neuronopathy and its clinical presentation.

With an acute or subacute course, an age over 60 years, an inflammatory CSF, or a marked pain point toward a seronegative paraneoplastic sensory neuronopathy (about 10% of cases), an FGF-PET scanner must be performed. If this is negative, one should consider infectious, dysimmune, or idiopathic sensory neuronopathy. If the progression is slow with predominant ataxia and few positive symptoms such as paresthesia and pain, three possibilities should be considered: (1) genetic sensory neuronopathy particularly if there are family cases and an early onset. Chronic cough may also point toward a genetic sensory neuronopathy (CANVAS). In these cases, genetic tests should be considered; (2) dysimmune sensory neuronopathy, particularly if the onset is asymmetrical or multifocal and the face or trunk are involved. Patients with this pattern should regularly be investigated for Sjögren syndrome or other autoimmune diseases that may appear years after the neuropathy onset; (3) finally, an important subgroup will remain unexplained, which as discussed probably conceals at least cases of dysimmune or genetic sensory neuronopathy.

Management

There is today no demonstrated curative treatment for any of the etiologies of sensory neuronopathies. So far genetic forms cannot be cured. Neuroprotective treatments to prevent chemotherapy-induced sensory neuronopathy are still wanted, although progress is ongoing in this field (45). Dysimmune sensory neuronopathy (including paraneoplastic sensory neuronopathy) are currently the most appropriate to benefit from ad hoc treatments. However, the main limitation is that to have chance to be efficacious treatments should be applied before the definitive degeneration of sensory neurons. A study based on the evolution of sensory action potential with time taken as a surrogate marker of sensory neuron cell death showed that the therapeutic window (ie, the time before which sensory action potentials are not recordable) is very short (04). It does not exceed three months in acute or subacute forms and is probably longer in chronic forms. Future therapeutic trials should take this crucial point into consideration. This also underlines the necessity of a diagnostic strategy and the availability biomarkers to reach as soon as possible a diagnosis of dysimmune sensory neuronopathy. Another concern is that there is no specific validated scale sensitive to change that could be used in such trial. A simple scale has been proposed that should be confirmed in a multicenter study (42).

Special considerations

Pregnancy

Immunosuppressive therapies are not recommended before and during pregnancy.

Anesthesia

In cases of associated autonomic neuropathy, hemodynamic and delayed abnormal respiratory, bowel, as well as bladder responses may occur during anesthetic and recovery periods.