Clinical manifestations

Presentation and course

In the majority of patients with paraneoplastic sensory neuronopathy (up to 86%), the neurologic symptoms are the presenting feature of the associated neoplasm, preceding discovery of the tumor by intervals ranging from several months up to 2 years or more (87; 28; 95; 69; 148; 173). The most prominent early symptoms are acute to subacute numbness and paresthesias, often patchy or asymmetric at onset, involving the arms but eventually involving all limbs (47; 33; 87; 28; 128; 100; 89; 66; 174). However, the clinical course has been reported to be more insidious in about 10% of cases (113). Burning dysesthesias and severe aching or lancinating pain occur in a majority of patients and may be the most distressing feature of the disease (78). Sensory involvement of the face, trunk, or abdomen may help point toward the diagnosis (93). Examination reveals severe sensory ataxia and predominant impairment of vibration sense and proprioception, although all sensory modalities are generally involved. Pseudoathetosis is common. Muscle stretch reflexes are diffusely hypoactive or absent. True weakness is rare and may be seen due to additional involvement of motor nerve roots, but patients with severely impaired proprioception may also have difficulty in sustaining effort during the motor examination (71). Most patients cannot walk unassisted due to pain and profound loss of proprioception. A minority of patients have prominent pain and mechanical hyperalgesia with at least partial preservation of large fiber sensibility and muscle stretch reflexes (106; 128). Dysautonomia is also common, with gastrointestinal dysmotility being a frequent manifestation (173).

In 2004, the Paraneoplastic Euronetwork consortium for paraneoplastic neurologic disorders suggested diagnostic criteria for paraneoplastic sensory neuronopathy, stating that a classical sensory neuronopathy should have the following characteristics: (1) subacute onset with a Rankin score of at least 3 before 12 weeks of evolution; (2) onset of numbness and often pain; (3) asymmetry of symptoms at onset; (4) arm involvement; (5) proprioceptive loss in areas affected; and (6) electrodiagnostic studies that demonstrate pronounced sensory fiber involvement and at least one absent sensory nerve action potential (68). In the updated 2021 diagnostic criteria for paraneoplastic neurologic syndromes, the clinical presentation of sensory neuronopathy is no longer referred to as a “classical paraneoplastic syndrome,” but it is now defined as a “high-risk phenotype” that frequently has a paraneoplastic etiology (71).

Many patients with paraneoplastic sensory neuronopathy have additional signs and symptoms that reflect a multifocal encephalomyeloneuritis. These include varied combinations of altered sensorium, confusion, memory loss, personality change, ocular dysmotility, dysarthria, dysphagia, cerebellar ataxia, corticospinal tract findings, asymmetric lower motor neuron weakness, intestinal pseudo-obstruction, or autonomic system degeneration (28; 37; 111; 95; 69; 148; 128). Conversely, patients with predominant limbic or cerebellar manifestations of paraneoplastic encephalomyeloneuritis often also have a component of sensory neuronopathy. Some patients with sensory neuronopathy develop concomitant Lambert-Eaton myasthenic syndrome, a component of motor neuropathy, or peripheral nerve microvasculitis presenting as mononeuritis multiplex (171; 55; 69).

The great majority of patients with paraneoplastic sensory neuronopathy or encephalomyeloneuritis deteriorate over weeks to months and then plateau at a level of severe disability with little to no improvement despite therapy (174). Less commonly, patients have stepwise deterioration (136) or exhibit a relentlessly progressive course leading to coma and death. Exceptional patients may have few, if any, overt central nervous system manifestations and a relatively indolent sensory neuronopathy that does not progress to neurologic disability even over several years (64).

Prognosis and complications

The course of paraneoplastic sensory neuronopathy in patients with small cell lung carcinoma is fairly stereotyped. Most patients deteriorate over several weeks or months and then stabilize at a level of severe neurologic disability (37; 69; 148; 03). Subsequent stepwise or gradual neurologic deterioration is less common and tends to occur in patients with less than complete response of the associated small cell lung cancer to treatment (95). Sudden, unexpected death is not rare and is presumed to be caused by acute dysautonomia. Conversely, a few patients have minimal central nervous system manifestations and a sensory neuronopathy that takes a relatively indolent course independent of any treatment (64). Paraneoplastic sensory neuronopathy or encephalomyelitis associated with anti-Hu (ANNA-1) antibodies may rarely spontaneously improve without any specific treatment (21; 139). The presence of paraneoplastic antibodies may be suggestive of an effective antitumor immune response, evident by the longer survival in patients with some autoantibodies in small cell lung carcinoma patients (62). One study that evaluated prognostic factors for paraneoplastic neurologic syndromes (including neuropathy) associated with small cell cancer found that autoantibody positivity or type of neurologic syndrome did not correlate with overall survival (173). The only exception was for paraneoplastic myelopathy, which was associated with reduced survival rates.

Clinical vignette

A 52-year-old right-handed woman without significant medical history other than cigarette smoking was found to have a possible right perihilar mass by chest x-ray done as part of a routine checkup. Bronchoscopy and transbronchial needle biopsy showed no definite tumor. Repeat chest x-ray 3 months later was unchanged, but she reported anorexia, constipation, and a 20 lb weight loss. Seven months after the initial chest x-ray, she developed numbness and painful dysesthesias extending from the elbows into all 10 fingers, much worse on the right side, with clumsiness of the right hand. She had similar numbness and dysesthesias from the left posterior thigh into the sole, with occasional "shooting" pains. CSF contained 12 white blood cell/mm3 and protein 88 mg/dL. The patient was prescribed trazodone for symptoms of depression, but after 1 week she became confused and had three generalized tonic-clonic seizures. A brain MR scan showed several small areas of abnormal signal in the hemispheric white matter. EEG showed left temporal lobe sharp waves. Repeat lumbar puncture showed 20 mononuclear white blood cells, protein 53 mg/dL, IgG index 2.5 (normal less than 0.7), and two oligoclonal bands. Phenytoin was begun and trazodone discontinued.

Over the next 3 weeks, the patient developed a weak voice, nasal regurgitation, a feeling of "swelling" in her throat, intermittent diplopia and vertigo, "wobbly" gait, and memory difficulty. On referral to another neurologist she had impaired short-term memory, diminished attention, and poor insight into her condition. Cranial exam showed vertical nystagmus on upward gaze, mild dysarthria, and myoclonic twitching of the larynx, lower lip, and left cervical strap muscles. Strength was normal except for 4+/5 bilateral proximal arm weakness and difficulty in sustaining effort with the right arm and hand. Muscle stretch reflexes were exaggerated in the left arm and barely able to be elicited in the other limbs. Plantar response was flexor bilaterally. Proprioception and vibratory sense were diminished in the right arm, and additionally there was stocking-glove impairment of light touch and pinprick. There was moderate ataxia and dysmetria of the right arm. Romberg test was positive, and gait was moderately ataxic.

Nerve conduction studies showed absent sensory nerve action potentials in both sural nerves and the right median nerve, and reduced sensory amplitudes in several other nerves. Chest CT scan demonstrated a right hilar and subcarinal mass. Biopsy of a newly palpable supraclavicular lymph node demonstrated small cell lung carcinoma. Staging workup revealed no other evident tumor. Serum and CSF contained high-titer anti-Hu (ANNA-1) antibodies.

The patient was given clonazepam, prednisone 60 mg per day, and monthly chemotherapy (cyclophosphamide + doxorubicin + vincristine). After three cycles of chemotherapy and 2 months of prednisone, tumor response was significant, but she reported worsening dysarthria, dysphagia, numbness and dysesthesias of all four limbs, and decreased gait stability. Repeat brain MR scan was unchanged. EMG testing showed no evidence for Lambert-Eaton myasthenic syndrome. Prednisone was tapered, and she received a course of six plasma exchanges. After 2 weeks she reported slight improvement in swallowing and gait stability; neurologic exam was unchanged. One month later she had another seizure and developed worsening diplopia, dysarthria, dysesthesias, and sensory gait ataxia (despite complete tumor remission after six cycles of chemotherapy). Prednisone was increased back to 60 mg per day, without clinical benefit. She died suddenly at home 2 months later.

Biological basis

Etiology and pathogenesis

The clinical features, electrophysiologic abnormalities, and neuropathology of paraneoplastic sensory neuronopathy indicate that the cell body of primary sensory neurons is the primary site of injury. The leading theory at present is that an autoimmune response initially directed against tumor cell antigens subsequently "spills over" to attack dorsal root ganglion neurons (and other neurons) (50; 58). The inflammatory infiltrates or antineuronal antibodies present in some patients indirectly support an autoimmune etiology, but the actual pathophysiologic mechanisms remain unknown.

Pathology. The neuropathologic hallmark of paraneoplastic sensory neuronopathy is severe dropout of primary sensory neurons in the dorsal root ganglia and gasserian ganglia (47; 33; 87; 88). The neuronal loss is diffuse but patchy and usually asymmetric, reflecting patients' clinical presentation. Large-diameter neurons are probably lost preferentially (126; 88). Remaining neurons occasionally show nonspecific degenerative changes. Nodules of Nageotte are often present. There is a highly variable degree of infiltration by T and B lymphocytes, plasma cells, and macrophages, often in a perivascular distribution (129; 168). Severe depletion of myelinated fibers is present in dorsal columns, posterior nerve roots, and peripheral sensory nerves, believed to be secondary to the loss of dorsal root ganglia neurons.

There are well-documented cases of sensory neuronopathy associated with small cell lung carcinoma in which the pathological abnormalities are restricted to the findings listed above (47; 33; 99). Some of the patients with sensory neuronopathy, however, have multifocal encephalomyeloneuritis, with patchy neuronal loss and variable inflammatory infiltrates in any or all areas of the nervous system including the cerebral hemispheres, limbic system, brainstem, cerebellum, gray matter of the spinal cord, dorsal root ganglia, and autonomic ganglia (33; 83; 69). Neuronal loss is accompanied by a variable degree of perivascular and leptomeningeal infiltration of mononuclear cells, including T and B lymphocytes and plasma cells (15). There is only a rough correlation between individual patients' clinical manifestations and the severity of neuronal loss and inflammatory changes in different sites seen at autopsy (35). Microvasculitis of peripheral nerves may rarely be superimposed on the encephalomyelitis (94; 171; 55).

Immunology. The theory that paraneoplastic sensory neuronopathy/encephalomyelitis is an autoimmune disorder is supported by the "inflammatory" histopathology and by the presence in some patients of circulating antineuronal autoantibodies. The most prevalent of these autoantibodies are polyclonal IgG anti-Hu antibodies (also called type 1 antineuronal nuclear antibodies (ANNA-1). Anti-Hu antibodies are present in up to 2% of cancer patients and were initially found in patients with small cell lung carcinoma and paraneoplastic sensory neuronopathy (65; 109). It has since been described in patients with various clinical manifestations of paraneoplastic encephalomyeloneuritis (37; 111; 95). Around 85% of patients with anti-Hu antibodies and paraneoplastic encephalomyeloneuritis have small cell lung carcinoma, with reports of patients with other tumors, including neuroblastoma, nonsmall cell lung carcinoma, or carcinoma of the breast, ovary, or prostate (12; 69; 141). In one series, sensory neuronopathy was the most common presentation in patients with anti-Hu antibodies, occurring in about 54% of cases (137).

Immunohistochemical studies utilizing anti-Hu antibodies stain the nuclei and, to a lesser degree, the cytoplasm of all neurons in the human brain, spinal cord, dorsal root ganglia, and myenteric plexus (65; 01). In immunoblots of human neuronal extracts, anti-Hu antibodies react with a group of closely spaced 35- to 40-kd proteins. Intrathecal synthesis of anti-Hu antibodies (out of proportion to the serum titer) is common (140) and is probably more prevalent among patients with clinically overt encephalomyelitis plus sensory neuronopathy than among patients with relatively "pure" sensory neuronopathy (162).

Anti-Hu antibodies react with a group of neuronal protein autoantigens, which are closely related structurally and antigenically but are encoded by different genes. The cloned protein autoantigens include HuD, HuB, HuC, and HuR (152; 98; 145). These proteins bind to AU-rich elements in the 3' untranslated region of mRNAs and regulate the stability of mRNA transcripts. The proteins thereby play a key role in the development, maturation, and maintenance of neurons (84; 18; 19). Multiple isoforms of the Hu proteins differ in their neuronal distribution and antigenicity (147).

Expression of one or more Hu autoantigens is common but is not universal among small cell lung carcinomas, including tumors from patients with paraneoplastic sensory neuronopathy or encephalomyelitis and anti-Hu antibodies as well as tumors from neurologically unaffected patients (36). There is marked heterogeneity in the level of expression of mRNA for individual Hu proteins among individual tumors (112; 49). No definite genetic mutations or antigenic differences have been identified between Hu autoantigens expressed by tumors from paraneoplastic sensory neuronopathy or encephalomyelitis patients and the proteins from neurologically unaffected small cell carcinoma patients (147).

At least a portion of the polyclonal anti-Hu antibody response in paraneoplastic sensory neuronopathy or encephalomyelitis is probably directed against one or more epitopes shared by the Hu autoantigens (52). These shared epitopes most likely reside in the portions of highest sequence homology among the autoantigens, namely, in the "RNA recognition motifs" that bind RNA. Differences in the fine specificity of the anti-Hu autoimmune response among different individuals may in part explain the clinical heterogeneity of patients with paraneoplastic sensory neuronopathy or encephalomyelitis.

To date, no fully successful experimental model exists for paraneoplastic sensory neuronopathy or encephalomyelitis. Passive transfer of human anti-Hu antibodies into mice fails to produce any histopathologic changes in the cerebellum, spinal cord, or dorsal root ganglia. In vitro incubation of patients' anti-Hu sera with human tumor cell lines or neuronal cultures has produced apoptosis or cytotoxicity in some laboratories but not in others (72; 85; 154). Animals actively immunized with recombinant Hu antigens (150; 149) or with HuD DNA vaccination (25) produce anti-Hu antibodies, but they do not develop the clinical or pathologic features of sensory neuronopathy or encephalomyelitis.

Currently it is believed that cellular immune effectors, and not anti-Hu antibodies, directly cause neuronal injury in paraneoplastic sensory neuronopathy or encephalomyelitis. The growing evidence for T lymphocyte-mediated neuronal injury includes: increased numbers of CD4+ lymphocytes, CD8+ T lymphocytes, regulatory T cells, and dendritic cells in the CSF of patients (42; 44); the presence of CD8+ T cells expressing cytotoxic proteins and clustered around neurons in brain and dorsal root ganglia (15; 17); stimulation of lymphocyte proliferation by recombinant HuD protein in vitro (14; 79); clonal expansion of certain T cell receptors in patients with paraneoplastic sensory neuronopathy or encephalomyelitis (167); and oligoclonal T cells in the blood and dorsal root ganglia of patients with anti-Hu antibodies (135). Circulating HuD-specific cytotoxic CD8+ T lymphocytes in affected patients were found in some studies (156; 155; 134; 143) but not in others (41; 43).

In a disproportionately high percentage of patients with paraneoplastic sensory neuronopathy or encephalomyelitis and anti-Hu antibodies, the associated small cell lung carcinoma is limited to the lung and mediastinum as compared with neurologically unaffected small cell cancer patients (67). Patients with anti-Hu antibody-associated sensory neuronopathy or encephalomyelitis appear to have a longer survival than patients with small cell lung cancer but no paraneoplastic syndrome (95). This is circumstantial evidence supporting an antitumor immune response, but an alternative explanation is that the occurrence of neurologic symptoms leads to early tumor diagnosis and treatment. There are reports of spontaneous regression of small cell cancer accompanied by or shortly preceding the onset of anti-Hu antibody-positive paraneoplastic sensory neuronopathy or encephalomyelitis (39; 59); whether the anti-Hu immune response is directly involved in this is unknown.

Some patients with paraneoplastic sensory neuronopathy or sensorimotor polyneuropathy, most of who have small cell lung carcinoma, have autoantibodies distinct from anti-Hu antibodies (Table 1).

Next to anti-Hu antibodies, anti-CV2/CRMP5 antibodies are the most common. The collapsin-response mediator proteins (CRMP) are found in the adult central and peripheral neurons in addition to small cell lung carcinomas (172). Unlike anti-Hu antibodies, anti-CV2/CRMP5 cross react with antigens on peripheral nerves and may demonstrate demyelination and axonal degradation with perivascular inflammation on biopsy (05; 54). The perivascular inflammation can encroach on the smallest nerve venules and capillaries, a hallmark of nonsystemic nerve vasculitis. Clinically, these patients present with a very painful axonal polyradiculoneuropathy, which is usually asymmetric (54). Patients may have concomitant cerebellar ataxia, myelopathy, or optic neuritis/retinitis. Electrodiagnostic studies show axonal or mixed axonal and demyelinating pattern and may mimic chronic inflammatory demyelinating polyneuropathy (CIDP). Small cell lung cancer was found in 75% of patients, and 15% had a thymoma, and symptoms typically preceded detection of underlying malignancy by months to years. Overall survival in these patients on average is better than those with anti-Hu (54).

In one series, contactin-1 IgG seropositivity was most commonly manifested as a sensory predominant demyelinating neuropathy (53). Sensory symptoms such as numbness, paresthesias, and sensory ataxia typically preceded any signs of weakness with a third of cases involving cranial nerves. Nerve conduction studies demonstrated prolonged distal latencies and slowed conduction velocities. Somatosensory evoked potentials in some cases were prolonged, suggesting proximal localization to the sensory nerve roots or dorsal root ganglia. Associated malignancies included thymomas, breast cancer, and plasmacytoma (53).

There are also case reports of sensory neuronopathy associated with inositol 1,4,5-triphosphate receptor type 1 (ITPR1) (92), anti-SOX1 (159), anti-Yo (153), anti-Ma2 (09), antiampiphysin (31), ANNA-3 (29), and anti-Trk paraneoplastic antibodies (121). Patients may have more than one type of antineuronal antibody (05; 131). Up to 16% of patients with small cell lung cancer and paraneoplastic sensory neuronopathy or encephalomyelitis have no detectable antineuronal antibodies (116). The clinical features of paraneoplastic sensory neuronopathy or encephalomyelitis in these “anti-Hu antibody-negative” patients do not differ from the spectrum of abnormalities in patients with anti-Hu antibodies.

Table 1. Antineuronal Antibodies in Paraneoplastic Sensory Neuronopathy and Sensorimotor Polyneuropathy

Antineuronal antibody	Associated tumors	Antibody reactivity
Anti-Hu (ANNA-1)	Small cell lung carcinoma, others	Pan-neuronal nuclei > cytoplasm; 35-40 kD RNA-binding proteins
Anti-CV2 (CRMP-5) (06; 172)	Small cell lung carcinoma, others	Cytoplasm of neurons and oligodendrocytes; 66 kD CV2 protein
Anti-amphiphysin (49)	Small cell lung carcinoma, breast carcinoma	Neuropil; synaptic vesicle-associated amphiphysin
ANNA-3 (29)	Small cell lung carcinoma	Purkinje cell and dentate neuronal nuclei; 170 kD protein
Anti-Ma	Lung, other carcinomas	Pan-neuronal nuclei and nucleoli; 37 kD Ma1 and 40 kD Ma2 proteins
Anti-Yo (115; 26)	Ovarian carcinoma, breast cancer	Neuronal cytoplasm

Epidemiology

More than 90% of reported patients with paraneoplastic sensory neuronopathy have small cell lung carcinoma (87; 28; 37; 69; 148; 04; 79). In a large prospective study of patients with small cell lung carcinoma, paraneoplastic sensory neuronopathy occurred in 1.9%, with or without other neurologic paraneoplastic manifestations (62). The syndrome can also occur in association with other neoplasms, including non-small cell lung carcinoma (95; 69; 138), breast carcinoma (28; 91; 141), ovarian carcinoma, (141), prostate carcinoma (10), uterine carcinoma (16), sarcoma (165), Merkel cell carcinoma (73), seminoma (120), thymoma (163), lymphoma (87; 144; 20; 132), neuroendocrine tumors (77), bladder carcinoma (166), and colon and gastric cancers (03).

Prevention

There are no known risk factors for developing a paraneoplastic sensory neuronopathy other than known risk factors for malignancy, such as smoking.

Differential diagnosis

The differential diagnosis of paraneoplastic sensory neuronopathy is fairly extensive and includes a variety of disorders that selectively affect sensory neurons or sensory nerve fibers (103; 23).

There are several well-documented patients with multifocal encephalomyeloneuritis, many with prominent sensory neuronopathy, whose clinical presentations and pathologic findings are indistinguishable from those of the paraneoplastic disorder, but in whom no tumor is ever discovered, even at autopsy (86; 38).

Subacute sensory neuronopathy can occur in patients with Sjögren syndrome (74; 119); in most of these patients the neurologic condition precedes recognition of Sjögren syndrome. Predominant loss of proprioception and vibratory sense (including the face and trunk), pseudoathetosis, areflexia, and sensory gait ataxia closely resemble the findings in paraneoplastic sensory neuronopathy. Biopsy of dorsal root ganglia has demonstrated neuronal loss and lymphocytic infiltration.

Sensory neuronopathy can occur in patients with gluten sensitivity. A few autopsied cases showed inflammation of the dorsal root ganglia and degeneration of the posterior columns of the spinal cord (81).

Pyridoxine toxicity is a well-documented cause of symmetric, pure sensory neuropathy (146). Most patients subacutely develop distal numbness and paresthesias and disproportionate impairment of vibratory sense and proprioception. Lhermitte symptom is often present, consistent with combined central and peripheral degeneration of large-diameter axons (02). Partial recovery occurs after the pyridoxine is stopped.

There are numerous reports of idiopathic sensory neuronopathy, sensory neuropathy, or ataxic neuropathy with the tempo of the disease ranging from acute and fulminant to chronic and slowly progressive (34; 170; 90). These patients share a clinical profile of numbness, paresthesias, dysesthesias (often asymmetric and involving the face and trunk), disproportionate impairment of proprioception and kinesthetic sensation, areflexia, preserved strength, and ataxic gait, all leading to severe disability. A few patients have superimposed autonomic insufficiency (82). Autopsy reports of neuronal loss in the dorsal root ganglia and autonomic ganglia with perivascular and perineuronal infiltration of T lymphocytes suggest a cell-mediated autoimmune process (82). Serum autoantibodies (distinct from anti-Hu antibodies) reacting with the cytoplasm of dorsal root ganglion neurons and Purkinje cells have been reported in several patients with nonparaneoplastic chronic sensory neuronopathy (122).

One described group of patients developed patchy, multifocal pain and loss of pinprick sensation in a non-length-dependent pattern, with sparing of large fiber sensation and no weakness or gait ataxia (60). None of the patients had anti-Hu antibodies or an identifiable neoplasm. The syndrome is believed but not proven to be a sensory ganglionopathy selectively affecting small nerve fibers.

Chronic sensory polyneuropathy may occur in patients with a serum IgM paraprotein reacting with GD1b and several other disialogangliosides (127; 57; 169; 08). Patients develop relapsing/remitting or steadily progressive numbness, paresthesias, areflexia, and sensory gait ataxia resembling paraneoplastic sensory neuronopathy. Some patients additionally have ophthalmoparesis and bulbar involvement, including respiratory insufficiency.

Acute demyelinating polyneuropathy may rarely present with severe sensory loss and ataxia with minimal weakness (40; 90); the relationship of this disorder to Guillain-Barré syndrome is unclear.

A small subset of patients with CIDP has a slowly progressive, "clinically pure" sensory neuropathy with preserved strength (125). Abnormalities in motor nerve conduction studies and electrophysiological evidence of primary demyelination distinguish these patients from those with sensory neuronopathy. In some patients the inflammatory/demyelinating process preferentially affects large myelinated fibers in the posterior nerve roots, resulting in a chronic progressive sensory ataxia (151).

Patients with HIV infection frequently develop a distal, symmetric sensory neuropathy with prominent burning dysesthesias (56). The pathogenesis of this neuropathy is probably multifactorial and includes neurotoxic effects of antiretroviral drugs (130).

Among patients with a known cancer diagnosis, the differential diagnosis of sensory neuropathy also includes neurotoxicity of chemotherapy drugs (51; 75). Sensory neuropathy is the major dose-limiting toxicity of cisplatin when given for treatment of several carcinomas. Patients develop subacute, distal numbness and paresthesias, dysesthesias, areflexia, and sensory ataxia. Lhermitte symptom or muscle cramps are common. The neuropathy often worsens and spreads proximally even after cisplatin is discontinued. Carboplatin is much less likely than cisplatin to cause sensory neuropathy, but severe sensory neuropathy has been reported in patients who received high-dose carboplatin after prior cisplatin-based regimens. Oxaliplatin, when given in high cumulative doses, can cause a sensory neuropathy that resembles cisplatin neuropathy (108; 101). Although checkpoint inhibitors have been implicated in a number of paraneoplastic neurologic syndromes including limbic encephalitis, they have not yet been reported in the development of neuropathy (66). However, there have been cases of anti-Hu neuronopathies worsening after initiation of checkpoint inhibitors that block the action of program cell death-1 (PD-1) protein (137).

Paclitaxel or docetaxel can cause a pure or predominantly sensory neuropathy (107; 102). Numbness and dysesthesias usually begin distally in the legs but can be asymmetric or involve the hands or face early in the course. Motor involvement can occur in the most severely affected patients. The neuropathy is probably related to the cumulative drug dose, as well as to the single-dose intensity. Patients who receive paclitaxel or docetaxel plus cisplatin may be especially likely to develop sensory neuropathy.

A mainly sensory polyneuropathy can be the dose-limiting toxicity in patients with multiple myeloma treated with thalidomide or with the proteasome inhibitor bortezomib (133; 07; 46).

The microtubule-stabilizing chemotherapy agent ixabepilone can cause a predominantly sensory polyneuropathy (161).

No clinical or laboratory features absolutely distinguish paraneoplastic sensory neuronopathy from the other disorders mentioned above. If present, signs or symptoms of multifocal encephalomyelitis or abnormal cerebrospinal fluid (especially pleocytosis, oligoclonal bands, or elevated IgG index) should raise suspicion of a paraneoplastic disorder, as these features are rare among patients with non-paraneoplastic conditions.

The presence of any of the antineuronal antibodies listed in Table 1 is highly suggestive of the presence of an underlying tumor, especially small cell lung carcinoma. Antibody assays for diagnosing paraneoplastic disorders do, however, have some false positives and false negatives. Anti-Hu antibodies have not been found in patients with dorsal root ganglionitis or sensory neuropathy secondary to Sjögren syndrome, plasma cell dyscrasias, chemotherapy toxicity, or a variety of other etiologies (27). There is a single report of anti-Hu antibodies in a patient with systemic lupus erythematosus and polyradiculoneuropathy (13). In a small percentage of patients with sensory neuronopathy/multifocal encephalomyelitis and high-titer anti-Hu antibodies, an associated tumor is never identified, either at autopsy or after several years of clinical follow-up (37; 24). The neurologic presentation and course of these anti-Hu-positive, nonparaneoplastic patients cannot be distinguished from those of patients with paraneoplastic sensory neuronopathy/multifocal encephalomyelitis (110). Depending on the methodology used, low serum titers of anti-Hu antibodies are detected in 15% to 40% of patients with small cell lung carcinoma, but without clinically overt sensory neuronopathy or encephalomyelitis (35; 52; 67; 164; 118).

A minority of patients with small cell lung carcinoma and sensory neuronopathy or encephalomyeloneuritis either have no detectable antineuronal antibodies (48), or have "atypical" autoantibodies with a pattern of reactivity distinct from anti-Hu or other well-characterized antibodies.

Diagnostic workup

The characteristic electrophysiologic profile of paraneoplastic sensory neuronopathy includes severely reduced amplitude or complete absence of sensory nerve potentials, with normal or only slightly reduced sensory nerve conduction velocities if a response is able to be elicited (87; 28; 06; 128). The pattern of abnormalities does not typically manifest in a distal to proximal gradient, differentiating it from more common length-dependent neuropathies (106). Most patients do show at least some electrophysiologic abnormalities in motor nerve conduction studies, with or without symptoms of a mixed sensorimotor polyneuropathy (22; 106; 124); motor conduction studies are almost always less affected than sensory nerve studies. Needle EMG may show denervation in patients with a component of motor neuropathy or encephalomyelitis and anterior horn cell involvement. Sural nerve biopsy may show nonspecific axonal degeneration and a reduced population of myelinated fibers (06; 128).

Spine MR imaging in patients with paraneoplastic sensory neuronopathy may show nonspecific T2-weighted signal abnormalities in the posterior columns, reflecting degeneration of the central projections of dorsal root ganglia neurons (105; 106). Cerebral spinal fluid may demonstrate a mild pleocytosis (30 to 40 white blood cells per cubic millimeter), a slightly elevated protein level (50 to 100 mg/dL), and/or an elevated IgG level. The pleocytosis will typically only last for the first few weeks to months, though the elevated IgG will persist (87; 37; 142; 93). Of note, normal cerebral spinal fluid findings do not rule out the diagnosis (63). Many patients additionally have oligoclonal bands, an elevated cerebrospinal fluid IgG index, or both.

A reasonable starting workup for patients with sensory neuropathy or neuronopathy includes lumbar puncture, nerve conduction studies and EMG, and serum assays for antineuronal antibodies, immunofixation, Sjögren syndrome A and Sjögren syndrome B, and HIV-1 serology. Screening for serum paraneoplastic antineuronal antibodies is done by a number of commercial or research laboratories. It is generally agreed that the preferred method for detection of antineuronal antibodies is a combination of immunocytochemical and immunoblotting studies (117; 131). Paraneoplastic antibodies are more likely to be detected in patients with higher grade tumors and when tumor density is higher (26). It should be recalled that a few patients with paraneoplastic sensory neuronopathy do not have antineuronal antibodies, and that a few patients with antineuronal antibodies are never found to have a tumor. This work-up can often be conducted in the outpatient clinic.

In patients who are found to have anti-Hu or any of the antineuronal antibodies listed in Table 1, the search for a tumor should focus on small cell lung cancer, with chest CT, MR scanning, or PET (30; 157). The concurrent presence of anti-SOX2 antibodies in patients with a neurologic presentation and serum anti-Hu antibodies further raises suspicion for an association with small cell lung carcinoma (61). There are reported patients with sensory neuronopathy or encephalomyelitis and antineuronal antibodies in whom chest CT or MR scans were unrevealing, but total body fluorodeoxyglucose-PET scans demonstrated a lesion proven to be small cell lung carcinoma (80; 114). PET scanning does have some false positives and negatives in this setting. If initial cancer screening, which includes CT and fluorodeoxyglucose-PET scans, are negative, then repeat screening should be performed every 4 to 6 months for 2 years, although some authors recommend up to 4 years (157; 71). The yield of “blind” bronchoscopy without a definite radiographic pulmonary lesion is low. It is not uncommon for patients' initial evaluation for an occult lung tumor to be unrevealing; in these patients the workup should be repeated every several months and should include workup for other tumors, including lymphoma and carcinoma of the breast or prostate.

Management

Empirical antitumor treatment without a histologic cancer diagnosis in patients with suspected paraneoplastic sensory neuronopathy or encephalomyelitis is not recommended. When the tumor is definitely diagnosed, it should be treated with the appropriate surgical, chemotherapeutic, or radiation measures. Unfortunately, patients often attain a remission or apparent cure of the underlying tumor but remain neurologically disabled.

In conjunction with treatment of the tumor, a trial of immunosuppressive therapy should be considered for patients with paraneoplastic sensory neuronopathy or encephalomyelitis. Immunosuppressive therapy should also be offered to patients with a clinical diagnosis of sensory neuronopathy and positive serum anti-Hu antibodies but no detectable neoplasm. Corticosteroids or IVIg are the most commonly used immunotherapies for paraneoplastic disorders. Patients generally receive an initially high dose of intravenous or oral steroid (prednisone, methylprednisolone, dexamethasone) followed by a slow or rapid taper. No clear evidence indicates the optimal drug, dose, route of administration, or schedule of corticosteroids. IVIg may be tried alone or in combination with corticosteroids. There is probably no indication for repeated courses of IVIg if the first course is not clearly effective. If corticosteroids and/or IVIg are not beneficial, "second-line" therapy options could include plasma exchange, cyclophosphamide, or azathioprine. Rituximab may also be considered due to its depleting effects on B-cell antigen presenters necessary for T-cell activation (104). An open label trial of sirolimus on 17 patients with anti-Hu associated paraneoplastic syndromes demonstrated treatment response in seven patients (45). In an open phase II trial evaluating the use of natalizumab in 15 patients with Anti-Hu associated sensory neuronopathy, objective improvement was reported in only 10% of patients although stability was seen in 60% of patients (11). These results may be driven, in part, by the short follow-up time of the trial. Symptomatic management of the patient’s pain may include the use of medication such as gabapentin, pregabalin, duloxetine, and venlafaxine (78). Major neurologic improvement is unlikely, but successful tumor treatment and early immunotherapy may increase the likelihood of at least some neurologic improvement (76).

There is theoretical concern that if the paraneoplastic disorder arises from an immune response directed against the tumor, attempts to treat the neurologic disorder with immunosuppression may adversely affect the evolution of the tumor. At this time, there is no definite evidence that patients given immunosuppressive treatment have a worse tumor outcome than those who are not (95).

Outcomes

Patients with paraneoplastic sensory neuronopathy rarely show significant neurologic improvement despite successful tumor treatment or a variety of immunosuppressive therapies, including prednisone, high-dose methylprednisolone, cyclophosphamide, IVIg, or plasmapheresis, whether alone or in combination (28; 37; 70; 160; 95; 96; 69; 148). It is difficult to interpret reports of "stabilization" of paraneoplastic sensory neuronopathy or encephalomyelitis with immunosuppression because the disease stabilizes spontaneously in most patients. Patients with a relatively short history of neurologic symptoms and who have successful treatment of the associated neoplasm may have a slightly better likelihood of neurologic stabilization or improvement (69; 24). Otherwise, nothing is distinctive about the clinical presentation of the few patients with paraneoplastic sensory neuronopathy and antineuronal antibodies who improve after immunosuppression, tumor treatment, or both (160; 123; 158; 32). Sirolimus suppresses cytokine-mediated T-lymphocyte proliferation and may, therefore, block cell-mediated neuronal injury. A small prospective study of sirolimus for patients with anti-Hu-associated paraneoplastic disorders, including sensory neuronopathy, did not show clearly better neurologic outcomes than those reported with other immunotherapies (45).

There are several potential explanations for the generally poor results of immunosuppressive therapy. It is possible that plasmapheresis or systemic immunosuppressive drugs do not adequately treat an autoimmune response "sequestered" within the CNS. Unfortunately, it is also more likely that at the time of diagnosis, many patients with paraneoplastic sensory neuronopathy or encephalomyelitis have already suffered irreversible neuronal damage or loss. This is supported by the typical monophasic clinical course in which patients deteriorate subacutely and then level off at a level of severe disability.

In contrast to the patients with paraneoplastic sensory neuronopathy and small cell lung carcinoma, several patients with Hodgkin disease have shown significant neurologic improvement following successful chemotherapy (144; 20; 132). There is evidence that the sensory neuropathy associated with Hodgkin disease is a demyelinating polyneuropathy rather than a dorsal root ganglionitis.