Subacute sensory neuropathy (SSN) is the most common paraneoplastic peripheral neuropathy (24 % of patients in a large European series, Giometto et al. 2010) and usually precedes diagnosis of the associated tumor, sometimes for as long as several years (Dyck et al. 1958; Dalmau et al. 1992; Chalk et al. 1992). Patients present with combinations of sensory ataxia, paresthesias, and limb pain, often with early involvement of the upper extremity, but with little or no weakness or wasting (although involvement of motor functions does not exclude SSN, Graus et al. 2004). Oki et al. (2007) divided paraneoplastic sensory neuropathy into patients with ataxia vs. those with pain which may reflect the size of targeted DRG neurons. Autonomic symptoms, such as urinary retention, orthostatic hypotension, or intestinal pseudo-obstruction, the last often a prominent early feature, may also occur in patients with subacute sensory neuronopathy (Oki et al. 2007). Autonomic features, present in 5 % of a large European series (Giometto et al. 2010) may herald a worse prognosis. A number of patients with autonomic involvement may have anti-Hu, anti-CV2/CRMP-5, or anti-Ach receptor antibodies (Lucchinetti et al. 1998; Yu et al. 2001; McKeon et al. 2009; Graus and Dalmau 2013). Progression occurs over days to weeks, often to a severity which precludes limb use. Concurrent paraneoplastic CNS involvement is common, including encephalopathy, cerebellar syndrome, or myelopathy. The CSF protein is typically elevated and the cell count usually normal. Electrical testing shows features of an axonal sensory neuropathy with normal or minimally altered motor conductions and EMG.

Survival from onset of neurological symptoms averages 2–3 years (Horwich et al. 1977), and death results as often from the neurological disease as from the underlying tumor (Dalmau et al. 1992). With a few exceptions, treatment appears to be ineffective, including that directed at the associated malignancy, but several patients with Hodgkin’s disease have experienced improvement with successful treatment of their malignancy (Smith 1992).

A clinically similar inflammatory sensory ganglionopathy can occur in the absence of systemic malignancy (Smith 1992). However, central nervous system features are absent, and the anti-Hu antibody (vide infra) is not detected.

Most cases of SSN result from damage to the spinal dorsal root ganglia (DRG, Fig. 16.1a). Autopsy studies have shown a loss of DRG neurons and focal chronic inflammatory infiltrates, chiefly CD8-positive cytotoxic T cells (Wanschitz et al. 1997; Ichimura et al. 1998), with resultant degeneration of the dorsal columns (Horwich et al. 1977). Large neurons may be preferentially affected (Ohnishi and Ogawa 1986). These patients should not undergo sural nerve biopsy, as their syndrome is characteristic, biopsy pathology is nonspecific, and serological tests for the associated onconeural antibodies are available. However, diseases with characteristic histology which might mimic SSN include sarcoidosis, paraprotein-associated neuropathies, and CIDP.

Nonetheless, pathological reports on a few patients are available (Lamarche and Vital 1987; Horwich et al. 1977; Dalmau et al. 1992; Dyck et al. 1958; Chalk et al. 1992). Sural nerve biopsy in patients with subacute sensory neuronopathy and marked sensory ataxia shows loss predominantly of large myelinated fibers, differing from preferential loss of small myelinated and unmyelinated axons in patients with pain predominant symptoms (Oki et al. 2007). Axon depletion in SSN ranges from mild to near total, with active degeneration a variably prominent feature (Fig. 16.1b, c). Axonal regenerative activity is usually minimal or absent, as expected in a neuronopathy (Lamarche and Vital 1987; Koike and Sobue 2008).

The majority of nerve biopsies reported in SSN have not shown inflammation, although one of our cases has shown significant sural inflammatory infiltration (Fig. 16.2a, b). However, prominent perivascular and vessel wall inflammation has occasionally been noted (Vallat et al. 1986; Plante-Bordeneuve et al. 1994) and, rarely, even vasculitis has been described (Johnson et al. 1979; Eggers et al. 1998). Such cases seem to reflect disease of the peripheral nerve rather than the spinal ganglion and may have a better prognosis in some patients (Plante-Bordeneuve et al. 1994) or may reflect a role for CV2/CRMP-5 which may occur with anti-Hu and be more targeted to peripheral nerve vs. dorsal root ganglia (vide infra).

The Paraneoplastic Neurological Syndrome Euronetwork (Giometto et al. 2010) has defined a number of antibodies (anti-Hu, Yo, CV2/CRMP-5, Ri, AChR, Ma2, Tr, and amphiphysin) which are thought to have a role in various paraneoplastic syndromes and may result in different clinical syndromes. Paraneoplastic SSN is most strongly associated with circulating antineuronal antibodies (anti-Hu, also called antineuronal nuclear antibody type 1, ANNA-1), which exhibits specificity for neuronal nuclear antigens (Smith 1992) and is most typically found in association with small cell lung cancer, much less frequently in other malignancies (Dalmau et al. 1992). It is thought that patients with anti-Hu antibodies alone most often present with subacute sensory neuronopathy, those with anti-CV2/CRMP-5 antibodies develop mixed axonal and demyelinating sensorimotor neuropathy, and those with both antibodies exhibit subacute sensory neuronopathy superimposed on demyelinating sensorimotor neuropathy (Antoine et al. 2001; Koike and Sobue 2013). Although patients with anti-Hu antibodies show a wide range of paraneoplastic neurological manifestations, including cerebellar ataxia, limbic encephalitis, myelopathy, Lambert–Eaton syndrome, and myopathy, peripheral neuropathy is the most common manifestation and present in 60–80 % of patients with SSN (Dalmau et al. 1992; Lucchinetti et al. 1998; Graus et al. 2001; Sillevis Smitt et al. 2002). The SSN DRG shown in Fig. 16.1c was accompanied by limbic encephalitis in the hippocampus (Fig. 16.1d). The anti-Hu antibody reacts with neoplasm antigens (Dalmau et al. 1992), leading to the assertion that the clinical syndrome is caused by antibodies produced against the tumor which cross-react with neuronal antigens. However, some of the lymphocytes infiltrating nervous tissue also appear to be sensitized to the Hu antigen, and cell-mediated damage may also play a role (Dalmau et al. 1992). In a previous study of 162 consecutive patients with anti-Hu antibodies, malignancy was present in 88 % of patients, and small cell lung cancer was subsequently identified in 93 % (Lucchinetti et al. 1998). CD8⁺ T cells in the DRG of SSN include both cytotoxic CD8⁺-positive T cells and atypical noncytotoxic CD8⁺-positive T cells with an uncertain dynamic (Roberts et al. 2009). It has been proposed that small cell lung cancer may evade surveillance by skewing tumor antigen-specific CD8⁺ T cells to an unusual noncytotoxic phenotype, providing an explanation for variation of clinicopathological features in anti-Hu antibody-associated paraneoplastic neurological syndromes (Roberts et al. 2009). Anti-CV2/CRMP-5 is also strongly associated with lung carcinoma (Honnorat et al. 1996; Yu et al. 2001), mostly small cell type, which was ultimately found in 77 % of patients with anti-CV2/CRMP-5 antibodies. Unlike anti-Hu antibodies, anti-CV2/CRMP-5 antibodies cross-react with antigens in the peripheral nerve trunk rather than DRG neurons (Antoine et al. 2001). Findings suggestive of demyelination and axonal degeneration have also been reported in biopsy specimens from patients with anti-CV2/CRMP-5 antibodies (Antoine et al. 2001). Somatic neuropathy is the most common manifestation in patients with anti-Hu and anti-CV2/CRMP-5 antibodies, while antiganglionic AchR antibody is associated with autonomic neuropathies (Graus et al. 2004). However, passive transfer of these antibodies or immunization has failed to reproduce the disease in animals (Sillevis Smitt et al. 1995).

A small proportion of patients with the antibody and the clinical syndrome have had no malignancy detected, even at autopsy (Dalmau et al. 1992). It is unclear whether the tumor was simply too small to be detected (Chartrand-Lefebvre et al. 1998; Lucchinetti et al. 1998) or the immune reaction to Hu antigen was caused by another factor. Fluorodeoxyglucose positron emission tomography (FDG-PET) may reveal the malignancy even if small (Rees et al. 2001; Younes-Mhenni et al. 2004; Titulaer et al. 2011).

Patients presenting with subacute sensory neuronopathy tend to show a better response to immunomodulatory treatment compared to patients with other classic paraneoplastic neurological syndromes (Uchuya et al. 1996; Keime-Guibert et al. 2000; Graus et al. 2001; Vernino et al. 2004).

Whereas SSN is a rare but characteristic paraneoplastic syndrome, a more nondescript sensorimotor neuropathy is seen in 5–16 % of cancer patients, with as many as 30–40 % showing subclinical electrophysiological abnormalities (McLeod 1993; Croft and Wilkinson 1965; Hawley et al. 1980; Walsh 1971; Graus et al. 1983; Paul et al. 1978). Small cell lung cancer is not as uniformly detected with these neuropathies as it is with SSN, and every type of malignancy has been associated with sensorimotor neuropathy, most often lung, breast, ovarian, stomach, and prostate cancers, as well as lymphoma (Croft and Wilkinson 1965; Walsh 1971; Dayan et al. 1965). Although the existence of this nonspecific sensorimotor “paraneoplastic” neuropathy is accepted, it is less precisely defined than SSN. Nonetheless, there is an association of patients with anti-CV2/CRMP-5 antibodies and the development of mixed axonal and demyelinating sensorimotor neuropathy, and those with anti-Hu and anti-CV2/CRMP-5 antibodies exhibit subacute sensory neuronopathy superimposed on demyelinating sensorimotor neuropathy (Antoine et al. 2001; Koike and Sobue 2013).

The paraneoplastic sensorimotor neuropathies are not homogenous, and classification is difficult (Corbo and Balmaceda 2001) because the pathophysiology of these entities is not understood and clinical features overlap. One approach that has been commonly used (Croft et al. 1967; McLeod 1993) distinguishes three groups on a clinical basis. Most frequently seen is a mild sensorimotor neuropathy that appears after the diagnosis of malignancy, sometimes called “mild terminal neuropathy”; the only abnormality may be hypoactivity of distal reflexes. The second group shows a more rapidly evolving and severe polyneuropathy that as often as not presents before the malignancy is diagnosed. The third and smallest group is composed of patients with a relapsing–remitting sensorimotor neuropathy. Regardless of where the patient falls into this classification, the neuropathic manifestations are typically sensorimotor, symmetrical, and predominantly distal, although proximal wasting and weakness can suggest a “neuromyopathy” (Croft and Wilkinson 1963). Motor neuron degeneration in the spinal cord at autopsy has been reported (Dalmau et al. 1992; Verma et al. 1996) and may be the cause of the motor component of paraneoplastic sensorimotor neuropathy in patients with anti-Hu-associated paraneoplastic neuropathy.

CSF protein is often elevated in the rapidly evolving and relapsing groups, but rarely in patients with “mild terminal neuropathy.” In this latter group, the electrophysiological assessment usually demonstrates axonal features, but in the more severe sensorimotor and the relapsing/remitting groups, a demyelinating or mixed pattern is often seen (Croft et al.1967; Paul et al. 1978; Graus et al. 1983).

As a rule, treatment of the tumor does little for the neuropathy, but exceptions have been reported, especially for lymphoma and seminoma (Evans and Kaufman 1971; Littler 1970; Enevoldson et al. 1990). Anecdotal reports indicate improvement with steroid or plasma exchange (Croft et al. 1967; Valbonesi et al. 1985).

Autopsy studies have not shown disease of the spinal ganglia in paraneoplastic sensorimotor neuropathy except when SSN was superimposed (Croft et al. 1967). Inflammatory infiltrates have also been sparse in proximal nerve trunks (Croft et al. 1967). The literature on the peripheral nerve pathology of paraneoplastic neuropathy is scanty, since patients with a mild “terminal” neuropathy are unlikely to undergo a nerve biopsy. Moreover, some authors have not distinguished between the various sensorimotor neuropathy groups.

In contrast to paraneoplastic subacute sensory neuropathy, the cause of paraneoplastic sensorimotor neuropathy remains more obscure. Although previously no tumor-related immunologic alteration or toxin has been found to affect peripheral nerve (McLeod 1993; Ongerboer de Visser et al. 1983), more recent studies have identified onconeural antibodies. Specifically, patients with anti-CV2/CRMP-5 antibodies are now known to develop mixed axonal and demyelinating sensorimotor neuropathy, and those with both antibodies exhibit subacute sensory neuronopathy superimposed on demyelinating sensorimotor neuropathy (Antoine et al. 2001; Koike and Sobue 2013). In some studies, the frequency of neuropathy was correlated with weight loss (Hawley et al. 1980; Graus et al. 1983; Hildebrand and Coers 1967), but this is not consistent even for “mild terminal neuropathy” (Croft et al. 1967).

In the rapidly evolving or relapsing paraneoplastic neuropathies, segmental demyelination and onion bulbs have been seen, although not usually to the extent seen in CIDP or other hypertrophic neuropathies. This suggests an important role for myelin injury. Indeed, CIDP may be “precipitated” by immunologic disturbances associated with the tumor (Sumi et al. 1983). However, segmental demyelination in paraneoplastic sensorimotor neuropathy may be secondary, as suggested by some teased fiber studies (Schlaepfer 1974; Lamarche and Vital 1987; Walsh 1971).

Some paraneoplastic sensorimotor neuropathies may be “dying-back” axonopathies. Intramuscular nerve axonal swelling has been described as paraneoplastic phenomenon by several authors (Barron and Heffner 1978; Brownell and Hughes 1975; Hildebrand and Coers 1967; Awad 1968) and may represent the earliest stages of a distal axonopathy as seen first in the longest motor fibers (Barron and Heffner 1978); similar changes are seen in typical toxic distal axonopathies (Spencer and Schaumburg 1977). However, the specificity and significance of this finding are suspect given its high frequency in normal intramuscular nerves and in neuropathies which are not believed to be due to a “dying-back” process (Alderson 1992).

The co-occurrence of vasculitis and malignancy is well recognized (Sanchez-Guerrero et al. 1990), but peripheral nerve involvement seems to be rare. Nonetheless, peripheral nerve necrotizing vasculitis has been described in association with malignancy. Patients may present with a sensorimotor polyneuropathy (Harati and Niakan 1986; Torvik and Berntzen 1968; Naka et al. 1991) or as pure sensory syndrome (Johnson et al. 1979). Deposition of immunoglobulin or complement in the vessel has not been detected (Johnson et al. 1979). Given the infrequency of this association, the possibility of a chance co-occurrence of peripheral nerve vasculitis and systemic malignancy must be considered.

Guillain–Barré syndrome has been infrequently reported in association with Hodgkin’s disease (Lisak et al. 1977; Ropper et al. 1991; Julien et al. 1980; Vital et al. 1990). No special features characterized the few cases studied pathologically, and typical macrophage-mediated demyelination has been observed (Julien et al. 1980).

Lymphomatous neuropathy is diagnosed when malignant lymphoma infiltrates and damages the peripheral nerve. This has been called neurolymphomatosis by some (Diaz-Arrastia et al. 1992), although others have taken nosologic exception (Guberman 1984; Baehring et al. 2003; Grisariu et al. 2010). Diaz-Arrastia et al. (1992) have recently reviewed the subject and examined a total of forty cases from the literature. Clearly the disorder is much more common than such numbers would indicate, as we have seen 4 cases in our own laboratory. About half of patients are not known to have systemic lymphoma at the time of diagnosis. Even at autopsy, 7 of 22 patients had lymphoma restricted to the nervous system alone, 2 had a focal lymphoma with dissemination to peripheral nerves only, and disseminated lymphoma was present in 13 (Diaz-Arrastia et al. 1992). The lymphoma is almost invariably of the non-Hodgkin (NHL) type (98 %), and 75 % of appropriately studied cases were of the B-cell malignancies (Diaz-Arrastia et al. 1992). Neuropathy is usually diffuse and sensorimotor in type and can evolve acutely mimicking Guillain–Barré syndrome, in a relapsing–remitting fashion, or most often progressing relentlessly over weeks to months (Diaz-Arrastia et al. 1992). One of the cases we have studied progressed slowly over 2 years. Pain may be a prominent feature. Electrophysiological testing can demonstrate axonal, demyelinating, or mixed features, and CSF examination often reveals elevated protein or CSF pleocytosis, with malignant cells in about half of cases (Diaz-Arrastia et al. 1992). Steroids, chemotherapy, or radiation therapy may be helpful. While the prognosis is usually poor (Diaz-Arrastia et al. 1992), in some patients, appropriate treatment may result in survival for many years (Ince et al. 1987).

Also intravascular lymphoma, formerly named “Tappeiner” disease or neoplastic angioendotheliomatosis, has been added to the concept of lymphomatous involvement (Glass et al. 1993). It can present as multiplex neuropathy, as Guillain–Barré syndrome (GBS) (Jiang et al. 2010), and as chronic inflammatory demyelinating polyneuropathy (CIDP) (Grisold et al. 2007; Briani et al. 2009, 2011). Intravascular lymphoma (Glass et al. 1993) is typically of B-cell origin and is thought to reflect lack of CD29 (β-1 integrin) and CD54 (ICAM-1) adhesion molecules on the neoplastic lymphoma cells interfering with transvascular migration (Ponzoni et al. 2000). Cauda equina syndrome, ascending polyradiculopathies (Levin and Lutz 1996; Patel et al. 2006; Jiang et al. 2010), multiple mononeuropathies (Roux et al. 1995), isolated mononeuropathies (Vital et al. 1989), symmetrical neuropathies (Oei et al. 2002), and muscle involvement (Fallon et al. 2002) have been observed (reviewed in Grisold et al. 2013).

Because only half of patients are known to have systemic lymphoma, neurolymphomatosis is one instance where nerve biopsy may be essential in making the diagnosis (Diaz-Arrastia et al. 1992). Even if malignancy is known to be present, infiltration of the nerve may be the only visible manifestation of recurrence after treatment (Shoenfelt et al. 1983; Krendel et al. 1991). Descriptions below are based on literature material (Barron et al. 1960, case 5; Gherardi et al. 1986; Guberman et al. 1978; Ince et al. 1987; Krendel et al. 1991; Thomas et al. 1990; Vital et al. 1990; Diaz-Arrastia et al. 1992; Zuber et al. 1987; Kohut 1946; Stack 1991; Atiq et al. 1992; Shoenfeld et al. 1983; Gold et al. 1988) and our experience with 4 cases – two at autopsy and two at nerve biopsy.

Autopsy reviews provide an inconsistent picture of the frequency of peripheral nerve involvement. In two autopsy series of Hodgkin’s disease, peripheral nerve tumor infiltration was detected in 0/30 and 5/30 patients, but far more often there were nonspecific nonneoplastic inflammatory infiltrates (Dickenman and Chason 1958; Jellinger and Radiaszkiewicz 1976). In the study of Grisariu et al. (2010), neurolymphomatosis was due to non-Hodgkin’s lymphoma in 90 % of cases and to acute leukemia in 10 %. In up to 26 % of the cases, it occurred as an initial manifestation of malignancy. We have found only one report of Hodgkin’s lymphoma causing a histologically proven infiltrative clinical neuropathy (Barron et al. 1960). In NHL the incidence of detected peripheral nerve infiltration by tumor cells in autopsy material varied from 0/20 to 41/98 (Dickenman and Chason 1958; Jellinger and Radiaszkiewicz 1976), but the quantity and location of peripheral nervous tissue studied was not clearly specified. A wide variety of NHL types have been present, including cutaneous and non-cutaneous T-cell lymphomas and high- and low-grade B-cell lymphomas. No particular subtype emerges as particularly likely to infiltrate peripheral nerve. Analysis of the literature is limited by small numbers and lack of the modern immunohistochemical and cytological techniques now routinely used to categorize lymphoma.