Introduction

Sensory neuronopathies (SN) are a subset of peripheral neuropathies that result from destruction of the dorsal root ganglion (DRG) and trigeminal ganglion sensory neurons. They go by many names in the literature, including dorsal root ganglionopathies and even ataxic neuropathies. These disorders can be broadly classified into acquired, or inherited and degenerative SN. An in-depth discussion of the inherited and degenerative disorders, which include Friedreich ataxia, POLG mutations, cerebellar ataxia, neuropathy, and vestibular areflexia syndrome (CANVAS), hereditary sensory and autonomic neuropathy type 1 (due to SPTLC1 mutations), and facial-onset sensory motor neuronopathy, is beyond the scope of this review [1,2,3,4,5]. The clinical presentation of these heritable and degenerative SN is distinct from that of acquired SN given their additional neurologic features. Regardless of the cause—acquired, inherited or degenerative—as a result of the rarity of these diseases, there is relatively little published information. There have been no blinded, placebo-controlled, clinical trials investigating the treatment of these conditions. Most publications are isolated case reports, retrospective case series, or reviews. We will expand on the various causes of acquired SN and their treatment strategies, with an emphasis on recent developments, followed by a recommended approach to diagnosis.

Anatomy

The DRG houses the cell bodies of the sensory pseudounipolar neurons. There are two subsets of neurons that make up the substance of the DRG. The larger cells give rise to Aβ and Aδ fibers. These carry proprioceptive and tactile sensation through the dorsal columns of the spinal cord. The smaller, unmyelinated C fibers carry thermal and nociceptive sensation through the spinothalamic tracts. The neurons of the DRG are susceptible to antibodies and toxins as a result of the loose blood–nerve barrier caused by fenestrated capillaries. The histopathologic findings in the DRG of inflammatory SN include reduced neuron density, satellite cell proliferation forming Nageotte nodules, which develop following degeneration of the ganglion cell bodies, and inflammatory infiltration (Fig. 1) [6].

Fig. 1
figure 1

Microphotography of the dorsal root ganglia. a Normal neurons. Hematoxylin and eosin (HE), ×200 magnification. b With evidence of neuronal loss and Nageotte nodules (arrows). HE, ×200 magnification. c Nageotte nodule detail (arrows). HE, ×600 magnification. d Mononuclear inflammatory infiltrate (arrows). HE, ×400 magnification. (Reprinted with permission from Colli et al. [6])

Full size image

Causes

Paraneoplastic Sensory Neuronopathies

Paraneoplastic syndromes associated with anti-Hu antibodies, and less often anti-collapsin-response mediator protein 5 (anti-CRMP-5)/CV2 antibodies, are one of the commonest causes of SN [7, 8]. The typical presentation is subacute sensory ataxia; however, patients can also present with concomitant motor neuropathy, limbic encephalitis, Lambert–Eaton myasthenia gravis, and cerebellar or brainstem involvement [7, 9, 10]. Autonomic features, including bilateral tonic pupils from effects on the ciliary ganglion, gastroparesis, and pseudo-obstruction, have also been reported [11,12,13,14]. In one case series, pain was a predominant feature in 80% of the patients in whom SN had been diagnosed [15]. Anti-Yo antibodies have been reported in a monomelic SN in a patient with invasive ductal carcinoma of the breast [16]. Amphiphysin antibodies have also been described associated with SN [17]. The association between anti-Hu paraneoplastic syndromes and small-cell lung cancer is widely known [18]. There are, however, a host of other malignancies with which anti-Hu antibodies have been linked, including uterine and other gynecologic tumors, testicular tumors, bladder tumors, prostate cancer, and even a rare case of hepatocellular carcinoma [19,20,21,22,23,24].

Pathologically, anti-Hu antibodies are generated by the immune system against Hu-expressing tumor cells but also destroy Hu-expressing neurons. They can penetrate at the level of the DRG, where the blood–nerve barrier is fenestrated and permeable. However, the role of anti-Hu antibodies in direct neuronal damage has yet to be elucidated. They are not thought to be directly responsible for the immune attack; rather it appears that this is primarily a CD8 cytotoxic T cell response [25, 26]. Anti-Hu antibodies react with HuD (ELAV-like neuron-specific RNA-binding protein 4) as well as HuC and HuB, which are neuron-specific proteins that bind messenger RNA. HuD is expressed by small-cell lung cancer cells, and likely initiates the autoimmune response in these patients [27, 28]. Tumors with anti-Hu antibody production express MHC type 1 molecules, further supporting the hypothesis that the anti-Hu response is T cell mediated [29, 30].

Treatment of paraneoplastic SN is challenging. The treatment strategy is based predominantly on expert opinion as there have been no randomized controlled clinical trials given the exceptionally rare nature of the disease. The mainstay of therapy lies in identifying and treating the underlying malignancy. In a retrospective study, tumor treatment was the only intervention that resulted in stabilization of the paraneoplastic syndrome in patients with anti-Hu antibodies [18]. The second principle of management is immunomodulatory treatment. We often initiate corticosteroid therapy (intravenously administered methylprednisolone) or intravenous immune globulin (IVIG) therapy while the malignancy is being evaluated. A limited window of opportunity for disease stabilization exists, and following the inflammatory reaction, there is irreversible neuronal damage, at which point therapeutic intervention is likely futile [31••, 32]. A number of immunosuppressant and immunomodulatory therapies have been used in uncontrolled trials, including corticosteroid therapy, IVIG therapy, plasma exchange, and rituximab therapy [33,34,35,36,37]. Cyclophosphamide may be beneficial in refractory disease when a tumor is not identified [32]. Most recently, de Jongste et al. [38] published a prospective, open-label study using sirolimus to treat anti-Hu paraneoplastic syndromes. Sirolimus is an immunosuppressive drug that inhibits activated T cells (cytokine-mediated T-cell proliferation). In this study of 17 patients with progressive anti-Hu antibody mediated paraneoplastic neurologic syndromes, nine had SN. Of the 17, ten had stabilization of their disease, six had progression of their disease, and one patient had improvement in function, as measured by the modified Rankin scale. Efficacy was not compared among the various neurologic syndromes [38]. These results support the use of immunomodulating therapies, but further study is needed to determine the use of one therapy versus another. The third principle of management is that of symptomatic treatment. Gabapentin, pregabalin, duloxetine, venlafaxine, and amitriptyline are all effective in treating neuropathic pain.

Autoimmune Causes

In addition to the paraneoplastic SN, the other large category of acquired SN is that of systemic autoimmune disease. The most commonly reported is Sjögren’s syndrome, and its association with SN has been identified since the 1980s [39]. Although less common than other neurologic complications of Sjögren’s syndrome, SN is more specific to Sjögren’s syndrome, and tends to be more disabling [40,41,42,43,44]. Patients usually have a sensory ataxia, areflexia, and autonomic dysfunction [45, 46]. Brainstem involvement, likely with disease affecting the ciliary ganglion, has also been reported [47]. Not much is known about the pathologic basis of SN in Sjögren’s syndrome, although there is loss of neuronal cell bodies, and T-cell-mediated infiltration in the DRG has been demonstrated [48]. In one autopsy of an 88-year-old woman with Sjögren’s syndrome associated SN, there were diminished numbers of DRG neurons in the cervical, thoracic, and lumbar segments, and diminished numbers of sympathetic ganglion cells across all segments [46]. CD8 T cell infiltrates were visualized.

Treatment strategies for Sjögren’s syndrome associated SN are based on case series and case reports, as there have been no randomized clinical trials. Successful response has been reported with the use of corticosteroids in association with mycophenolate mofetil [41•]. Other treatment strategies include plasma exchange, rituximab therapy, cyclophosphamide therapy, and azathioprine therapy [49,50,51,52]. IVIG therapy has had variable reports of success, with some studies suggesting that SN is less responsive to IVIG than are small-fiber sensory or sensorimotor neuropathies [53]. Others found that the long-term use of IVIG in ataxic SN with Sjögren’s syndrome led to abatement of neurologic symptoms [54]. One case report by Caroyer et al. [55] demonstrated clinical and electrophysiologic improvement in the condition of a patient with SN due to Sjögren’s syndrome after 12 weeks of therapy with infliximab, and the improvements remained after 48 months. A recent study by Goodman [56] suggests that autonomic symptoms related to Sjögren’s syndrome may respond to repeated, ongoing, or adjunctive immunomodulation.

SN have also been identified in celiac disease, with or even without enteropathy, although this is somewhat controversial [57]. In celiac disease, the autoimmune response to gliadin results in small bowel inflammation, which resolves with removal of gluten from the diet [58]. Autopsy and biopsy data in these patients have demonstrated humoral immune activation in the peripheral nerves and DRG as well as other portions of the nervous system [57]. The commonest neurologic complication of gluten sensitivity is cerebellar ataxia, termed gluten ataxia, but various forms of peripheral neuropathy, including cases of SN, have also been reported [57, 59, 60]. In the hallmark article from 2010 by Hadjivassiliou et al. [57], 53 of 409 patients (13%) with chronic neuropathies were found to have SN [57]. SN was defined as patchy sensory loss with or without sensory ataxia, which was supported by electrodiagnostic evidence of a patchy, non-length-dependent sensory fiber involvement. Seventeen percent of the SN patients had serologic evidence of gluten sensitivity as determined by antigliadin antibodies. Of the 17 patients with SN and gluten sensitivity, only seven had enteropathy on biopsy ,and the rate of false positive results for antibodies was high. In all patients, cerebellar ataxia (gluten ataxia) had been excluded. Four patients demonstrated length-dependent sensory fiber involvement without motor involvement. Many would classify this as a length-dependent sensory axonopathy as opposed to SN. It has also been reported that celiac patients have a small-fiber neuronopathy [61]. In 2014, McKeon et al. [62] investigated the association of gliadin antibody positivity (without celiac disease) and neurologic dysfunction (including cerebellar ataxia, neuropathy, and other presentations). They determined that in most gliadin antibody positive patients without celiac disease, alternative causes of neurologic dysfunction, including nutritional deficiency and coexisting autoimmunity, could result in the neurologic manifestations. Reports of treatment response differ, with some neurologic symptoms stabilizing [57] or abating with initiation of a gluten-free diet [57, 61]. However, there is some suggestion that initiation of a gluten-free diet long after onset of neurologic symptoms does not produce significant abatement of symptoms [63].

Many other autoimmune diseases have been linked to SN. There are several case reports of SN in patients with autoimmune hepatitis [64,65,66]. Unfortunately, the patients’ neurologic symptoms did not significantly respond to immunomodulating therapies. Unlike in Sjögren’s syndrome, the association between SN and systemic lupus erythematosus is quite rare, with only a few case reports [67, 68]. A newly described association is that of livedoid vasculopathy and SN [69]. Livedoid vasculopathy is a cutaneous skin disorder that results in purpuric lesions that progress to ulcerations of the limbs. Healing of the ulcerations results in atrophie blanche (white atrophied scars). There is a report of a 32-year-old patient with livedoid vasculopathy who developed SN, initially presenting with gait ataxia and dysesthesia [69]. Alternative causes were excluded, and she was treated with mycophenolate mofetil and prednisolone. When the SN progressed, treatment with rituximab was added, which resulted in clinical and electrodiagnostic stabilization.

Anti-fibroblast growth factor receptor 3 (anti-FGFR3) antibodies have recently been associated with SN [70••]. In 2015, Antoine et al. [70••] studied 106 patients with pure sensory neuropathy, 72 of whom fulfilled the diagnostic criteria for SN. Anti-FGFR3 antibodies were identified in 17 patients (16 with a sensory neuropathy and 1 with systemic lupus erythematosus and no reported neuropathy). Antoine et al. proposed that the target of the antibodies is likely the DRG given that 87% of the antibody-positive patients had a non-length-dependent pattern of sensory disturbance and 82% fulfilled the diagnostic criteria for probable SN. The characteristic features include severe pain and trigeminal nerve involvement. One patient had severe dysautonomia. Nerve biopsy demonstrated fiber loss without regenerating clusters, supporting an SN over an axonopathy. Further, electromyography results were similar to those in SN. Antoine et al. reported that anti-FGFR3 antibodies were present in 19% of patients with idiopathic SN or SN associated with another autoimmune disorder in their series. These results suggest that identification of anti-FGFR3 antibodies may aid in diagnosis of SN.

Toxic Causes

Pyridoxine (vitamin B6)-induced SN was first described in the 1980s, and many cases have been reported since then [71, 72]. Pyridoxine exerts a dose-dependent effect on the DRG, with irreversible damage at higher doses despite cessation of intake of the vitamin [73]. In humans, SN has been reported with dosages anywhere from 200 to 6000 mg per day [72, 74]. Patients typically experience a severe sensory ataxia due to necrosis of the DRG and degeneration of their central and peripheral projections [75]. The exact mechanism of this necrosis has yet to be determined.

Another well-described toxic cause of SN is the platinum-based chemotherapeutic agents, including cisplatin, carboplatin, and oxaliplatin [76]. Cisplatin and oxaliplatin are considered more neurotoxic than carboplatin, with neurotoxicity reported in up to 90% of patients treated with oxaliplatin [77]. The effect is dose dependent, and with cisplatin, a cumulative dose higher than 300 mg/m2 has been demonstrated to result in electrophysiologic evidence of damage to large-fiber neurons, leading to SN-related symptoms [78, 79]. Platinum-containing drugs are known to result in peripheral nervous system toxicity through nuclear and mitochondrial DNA damage, inducing oxidative stress, and disturbing ion channels [80]. Cisplatin, in particular, is known to induce apoptosis in the DRG by binding to nuclear DNA and mitochondrial DNA, resulting in DNA damage and activation of p53 and Bax-mediated apoptosis [81]. Recent work by Maj et al. [82] suggests that prevention of the p53 accumulation with small-molecule pifithrin-μ may be neuroprotective. The “coasting effect” refers to the emergence of a symptomatic neuropathy up to 2 months after cessation of the chemotherapy and clearance of the drug [76]. For this reason, it is important to stop use of the offending agent at the first sign of neuronal dysfunction.

Another potential toxic cause has recently been described in the literature. Novak et al. [83] reported on the association between a non-length-dependent, subclinical, small-fiber neuropathy and statin use. In this study, the skin and sweat gland biopsy specimens of 80 statin users were compared with those of age-matched controls. Novak et al. found that compared with the biopsy specimens of non-statin users, the biopsy specimens of statin users had significantly reduced epidermal and sweat gland nerve fiber densities at the thigh but not at the calf. Of importance, there were no significant differences between the two groups with regard to autonomic testing scores, or subjective pain or numbness scale scores. They concluded that this non-length-dependent process represents a new association between sensory and autonomic ganglionopathy and statin use.

Infectious Causes

Several infectious causes have been implicated in the development of SN. The commonest is HIV/AIDS, but others include Epstein–Barr virus, varicella–zoster virus, and human T-lymphotropic virus [84,85,86,87,88,89]. Other infectious associations include leprosy, particularly in countries such as India, where the disease is commoner [90]. Recently, Chiu et al. [91] reported on a case of a 3-year-old girl in whom enterovirus infection had been diagnosed [91]. Nine days later she developed dorsal column dysfunction and absent sensory nerve action potentials (SNAP) on nerve conduction studies. This is the youngest reported case of SN, and the only case associated with enterovirus to date.

Idiopathic Causes

Despite extensive workup, about half of all SN are deemed to be idiopathic [86, 92]. This is obviously a diagnosis of exclusion, although it is thought that these presentations are likely of autoimmune or possibly toxic cause [93]. Unlike most other subacute presentations described thus far, idiopathic SN are typically indolent and slowly progressive [92].

Diagnostic Evaluation

In 2009, Camdessanché et al. [94] proposed a set of diagnostic criteria for SN on the basis of a retrospective of a study population so as to help distinguish SN from other sensory neuropathies [94]. These criteria were later validated in a study of 210 patients at 15 referral centers [95•]. In this validation study, the SN criteria were 90.3% sensitive and 85.2% specific, and had a positive predictive value of 91.9% and a negative predictive value of 82.5%. The criteria were validated against expert diagnosis as opposed to biopsy of DRG. See Table 1 for the score form for the SN diagnostic criteria.

Table 1 Causes of acquired sensory neuronopathies and recommended evaluation
Full size table

When workup is being performed on a patient with suspected SN, electrophysiologic testing can be the most helpful study to further support the diagnosis, and is strongly recommended. Non-length-dependent sensory abnormalities are the hallmark, and often the upper extremities are affected disproportionately [86]. These abnormalities are typically widespread and asymmetric [96]. Although classically motor nerve conduction study findings are normal, motor nerve conduction study abnormalities have been reported for most causes [45, 64, 87, 92, 97, 98]. Needle electromyography is remarkable for incomplete activation of muscles due to loss of sensory input and the resultant incomplete interference pattern [99]. Recently, serial nerve conduction study findings were studied in 86 inflammatory SN patients [31••]. There was a monthly reduction of SNAP amplitudes that worsened in the first 2 months, then slowed down after 7 months, and finally stabilized after 10 months. The decline in SNAP amplitudes correlated with disability progression and lymphocytic reaction in the cerebrospinal fluid (CSF). On the basis of this observation, the study authors proposed that treatment should be initiated within the first 8 months if possible. The authors recommend treatment within 2 months of symptom onset during which the inflammatory reaction and decline in sensory response amplitude are at their greatest. The radial and ulnar nerves seemed to be the most sensitive to change.

Depending on the specific clinical picture and history of exposures, serum and CSF analysis can aid in the diagnosis of the specific autoimmune or toxic causes. In general, we recommend the study of the following for initial diagnostic evaluation: anti-Hu and anti-CRMP-5/CV2 antibodies, antinuclear antibodies, anti-SSA and anti-SSB antibodies, vitamin B6 level (if there is a history of supplementation), and HIV. Should the study findings be unremarkable, then a more extensive workup is warranted, including additional laboratory and imaging studies as outlined in Fig. 2. In the case of a paraneoplastic syndrome, the sensitivity and specificity of serum anti-Hu antibodies are 82% and 99% respectively [100]. Some patients with anti-Hu antibodies also have anti-CRMP-5/CV2 antibodies, and they may have a mix of SN and superimposed demyelinating sensorimotor neuropathy. Anti-CRMP-5/CV2 antibodies in isolation are associated with a mixed axonal and demyelinating polyneuropathy [8]. Notably, nearly 20% of paraneoplastic sensory neuropathies are seronegative [100]. In this case, characteristic CSF sampling may be useful, and demonstrates pleocytosis, elevated protein levels, and oligoclonal bands [18, 94, 101]. Importantly, neurologic symptoms are often the presenting symptom of a malignancy, and precede the cancer diagnosis by 3–8 months, and care should be taken to look for evidence of malignancy if a paraneoplastic syndrome is diagnosed [10, 18, 24, 102]. A chest computed tomography scan should be performed, and even fluorodeoxyglucose positron emission tomography should be considered if the initial imaging findings are negative and clinical suspicion remains high [103, 104]. If underlying malignancy is not found and the patient harbors paraneoplastic antibodies, patients should undergo repeated screening in 3–6 months, and then every 6 months thereafter for 4 years [105].

Fig. 2
figure 2

Score form for the diagnosis of sensory neuronopathy. MRI magnetic resonance imaging. (Modified with permission from Camdessanché et al. [94])

Full size image

Magnetic resonance imaging (MRI) of the spinal cord can also be helpful in the diagnosis of an SN as T2 hyperintensity in the dorsal columns may be observed [106]. Multiple-echo data image combination (MEDIC) and turbo inversion recovery magnitude (TIRM) imaging techniques have recently been described to reveal characteristic findings in SN. MEDIC imaging demonstrates higher signal intensity in the DRG and posterior columns. Conversely, TIRM imaging demonstrated a smaller spinal cord area and decreased nerve root diameter than in healthy controls [107]. Diffusion tensor imaging (DTI) has been investigated as a tool to differentiate SN patients from diabetic sensorimotor polyneuropathy patients and healthy individuals. In this study, MRI findings were compared among 28 patients with SN, 20 healthy controls, and 14 patients with diabetic distal sensorimotor polyneuropathy. It was found that DTI abnormalities in SN patients preceded the T2 hyperintensity in the dorsal columns on MRI, and the study authors proposed that it may be more sensitive. The study authors hypothesized that the DTI abnormalities are secondary to axonal loss and gliosis of the cuneate and gracile fasciculi, representing damage of the central projections in SN. They concluded that DTI imaging could be an earlier, noninvasive, in vivo test used in the diagnosis of SN and could also help differentiate between SN and distal neuropathies [108].

Depending on the likely underlying cause, tissue biopsy may also a useful diagnostic tool in SN. In patients with suspected Sjögren’s syndrome associated SN, a lip or salivary gland biopsy can be diagnostic should the test for autoantibodies be negative [109]. In paraneoplastic syndromes, tumor biopsy is required to confirm the pathologic features of the underlying malignancy. In patients with small-fiber-predominant SN (as in those with celiac disease), a skin biopsy may demonstrate a non-length-dependent pattern of nerve loss. The only definitive way to confirm DRG disease is by biopsy of the DRG, which is an invasive, traumatic procedure, which we do not recommend.

Conclusion

Acquired SN are a rare group of disorders that present with a non-length-dependent sensory neuropathy and early-onset ataxia. The prompt recognition and diagnostic evaluation of SN is important as an underlying malignancy is often discovered. Additionally, early identification of SN results in more immediate initiation of treatment, which has significant prognostic implications. In the future, further identification of associated autoantibodies may facilitate identification of this uncommon subset of neuropathies and may result in more successful, focused therapeutic interventions.