Clinical manifestations

Presentation and course

Paraneoplastic disorders can affect any part of the central or peripheral nervous system (13; 12). Many patients can be grouped into a recognizable clinical syndrome predominantly affecting 1 anatomic location or system (Table 1), such as limbic encephalitis or cerebellar degeneration. Other patients do not neatly fit into this scheme, as they have signs and symptoms of more than 1 syndrome. One view is that most, if not all, paraneoplastic disorders are subsets of a diffuse and multifocal "paraneoplastic encephalomyelitis" or "paraneoplastic encephalomyeloneuritis" in which individual patients may show predominant involvement of particular parts of the nervous system (15).

Table 1. Neurologic Paraneoplastic Disorders

Central nervous system	Peripheral nervous system
Multifocal encephalomyelitis Cerebellar degeneration Limbic encephalitis Opsoclonus-myoclonus Extrapyramidal syndrome Brainstem encephalitis Myelopathy Motor neuron disease Stiff-person syndrome Epilepsy Optic neuritis Retinal degeneration	Sensory neuronopathy Nerve vasculitis Sensorimotor polyneuropathy Motor neuropathy Neuromyotonia Autonomic insufficiency Lambert-Eaton syndrome Myasthenia gravis Polymyositis/dermatomyositis Necrotizing myopathy

Prognosis and complications

The neurologic outcome of patients with paraneoplastic syndromes varies considerably among different disorders. Some patients have major neurologic improvement solely with treatment of the associated neoplasm; this possibility places an obvious premium on prompt recognition of the paraneoplastic syndrome. Unfortunately, many patients are left with severe and permanent neurologic disability despite remission of the associated tumor. The success of tumor therapy or immunotherapy spans a spectrum: patients with Lambert-Eaton myasthenic syndrome generally have a gratifying response (36), whereas the great majority of patients with paraneoplastic encephalomyelitis or cerebellar degeneration are left with severe permanent neurologic disability despite aggressive treatment. The outcome for patients with paraneoplastic opsoclonus, limbic encephalopathy, or stiff-person syndrome usually lies somewhere between these extremes.

Clinical vignette

A 34-year-old woman presented with several weeks of worsening cognitive problems, unusual behaviors, and hallucinations, culminating in catatonia. Although her EEG and MRI of the brain with contrast showed no overt abnormalities, her cerebrospinal fluid analysis detected lymphocytic pleocytosis (11 cells), normal protein, and glucose. Her clinical picture was concerning specifically for autoimmune encephalitis due to anti-NDMAR antibody, and the search for an underlying tumor revealed a small ovarian teratoma. Intravenous steroids and plasma exchange were administered, and the tumor was excised. Serum autoimmune encephalitis panel was negative, but cerebrospinal fluid autoimmune encephalitis panel showed anti-NMDAR antibody positivity. The patient improved over the next several weeks. She subsequently received 2 infusions of anti-CD20 monoclonal antibody rituximab to decrease the risk of relapses. Over the next 6 months, she returned to her baseline functional level, with only mild neurocognitive sequelae detectable on neuropsychological testing.

Biological basis

Etiology and pathogenesis

Most neurologic paraneoplastic disorders are believed to be autoimmune diseases, although, with some exceptions, the exact immunopathogenic mechanisms remain unclear. For most paraneoplastic disorders, it is postulated that tumor cells express "onconeural" antigens, which are identical or antigenically related to molecules whose expression is normally restricted to neurons. In rare instances, an autoimmune response initially arising against the tumor antigen(s) "spills over" to attack neurons expressing the same or related antigens, leading to clinical neurologic disease.

Since the mid-1980s, there has been a steadily growing list of antineuronal antibodies identified in the sera of patients with paraneoplastic disorders (Table 2) (29; 06; 22). Some paraneoplastic antibodies have selective neuronal reactivity and are found only in patients with a particular clinical syndrome. Examples include antirecoverin antibodies in patients with retinal degeneration and anti-Yo (PCA1) or anti-Tr antibodies in patients with cerebellar degeneration. Most paraneoplastic autoantibodies show a more widespread or pan-neuronal reactivity and are associated with a variety of clinical neurologic syndromes or with multifocal encephalomyelitis (eg, anti-Hu [ANNA1] and anti-CV2 [CRMP5]). Patients (most frequently in the setting of small cell lung cancer) may have more than 1 type of autoantibody (20). The high degree of specificity of an autoantibody for a particular neurologic syndrome does not in itself prove that the antibodies are pathogenic.

The neuronal molecular targets of most autoantibodies have been characterized. Many of the initially discovered onconeural antibodies (eg, anti-Hu and anti-Yo) react with intracellular neuronal protein antigens, whereas most antibodies react with cell surface receptors, channels, or protein antigens associated with synaptic complexes (06; 22). Many of the protein antigens reacting with antineuronal antibodies are known to be expressed by the tumors from affected patients, providing circumstantial evidence supporting the general theory of an autoimmune response arising against shared "onconeural" antigens.

The proven or postulated immunopathogenic mechanisms for neurologic paraneoplastic disorders fall into 4 main categories:

(1) Autoantibodies against shared onconeural antigens are the direct cause of neurologic disease. For example:

	• Lambert-Eaton myasthenic syndrome is caused by antibodies that bind to and downregulate voltage-gated calcium channels at the presynaptic neuromuscular junction, leading to reduction in the quantal release of acetylcholine by a nerve impulse.
	• Antibodies against the Caspr2 protein associated with voltage-gated potassium channels in patients with neuromyotonia are believed to cause prolonged motor neuron depolarization, leading to abnormal spontaneous muscle activity.
	• Antibodies against synaptic receptors or trans-synaptic protein complexes (eg, NMDA receptors, glutamate receptors, GABAB receptors, or the LGI1 protein) may cause neuronal injury or dysfunction in patients with limbic encephalitis (21; 23; 24; 19; 30).
	• Antibodies against voltage-gated calcium channels or glutamate receptors may directly mediate Purkinje cell injury in some patients with paraneoplastic cerebellar degeneration (25).
	• Antiamphiphysin antibodies may directly contribute to causing paraneoplastic stiff-person syndrome by blocking presynaptic GABAergic inhibition (11).
	• Anti-recoverin antibodies probably exert a direct cytotoxic effect in patients with paraneoplastic retinal degeneration, perhaps in concert with cellular immune effectors.

(2) A cellular immune reaction against onconeural antigens is the main cause of neuronal injury. This is probably true for most patients with paraneoplastic cerebellar degeneration, for patients with multifocal encephalomyelitis or sensory neuronopathy associated with small cell lung cancer, and for at least some patients with paraneoplastic limbic encephalitis or opsoclonus-myoclonus. For these disorders, it is postulated that onconeural antigens released by apoptotic tumor cells are presented to T lymphocytes in draining peripheral lymph nodes, initiating a Th1 helper response that eventually gains access to the CNS and attacks neurons expressing the antigens. Evidence to support cell-mediated neuronal injury includes the presence of cytotoxic T lymphocytes in apposition to neurons in some patients (29; 01). Patients may also develop antineuronal antibodies (eg, anti-Yo, anti-Hu, and anti-Ma2. In vitro evidence suggests that anti-Hu and anti-Yo antibodies can exert a direct cytotoxic effect on neurons, but this effect is believed to be less important than that of cellular immune effectors in causing clinical disease. To date, there exist no fully successful animal models that reproduce cell-mediated paraneoplastic disorders in the CNS.

(3) Neoplastic plasma cells produce monoclonal paraproteins (immunoglobulins), which react with peripheral nerve antigens and cause neuropathy, eg, sensorimotor neuropathy associated with antimyelin-associated glycoprotein (anti-MAG) antibodies or sensory ataxic neuropathy associated with antidisialosyl ganglioside antibodies.

(4) Disorders that arise from other or poorly understood immune mechanisms, such as paraneoplastic necrotizing myelopathy, vasculitic neuropathy, and neuropathy associated with osteosclerotic myeloma and POEMS syndrome (34).

Table 2. Paraneoplastic Disorders and Autoantibodies

Clinical syndrome	Associated tumors	Autoantibodies*
Multifocal encephalomyelitis	SCLC, thymoma, testicular seminoma, others	anti-Hu, anti-CV2, anti-LGI1, anti-Ma1, anti-amphiphysin, anti-Ri (ANNA2), ANNA-3, Kelch-like protein (KLHL11)
Limbic encephalitis	SCLC	anti-Hu, anti-CV2, PCA-2, ANNA-3, anti-amphiphysin, anti- LGI1, anti-VGCC, anti-Zic4, anti-mGluR5, anti-GABABR, anti-AMPAR, anti-GAD, anti-Ri, anti-GFAP
	Testicular, breast	anti-Ma2, anti-mGluR1/2
	Thymoma	anti- LGI1, anti-CV2, anti-mGluR5, anti-AMPAR
	Hodgkin lymphoma	anti-mGluR5
	Ovarian teratoma, testicular	anti-NMDAR
Cerebellar degeneration	Breast, ovarian, testicular seminoma, others	anti-Yo, anti-Ma1, anti-Ri, Kelch-like protein (KLHL11)
	SCLC, others	anti-Hu, anti-CV2, PCA-2, ANNA-3, antiamphiphysin, anti-VGCC, anti-Ri, anti-Zic4, anti-GAD, anti-GABABR, anti-PKC
	Lymphoma, others	anti-Tr, anti-mGluR1
Opsoclonus-myoclonus	Breast, ovarian	anti-Ri, anti-Yo, antiamphiphysin
	SCLC	anti-Hu, anti-Ri, anti-CV2, antiamphiphysin, anti-VGCC
	Neuroblastoma	anti-Hu, others
	Testicular, others	anti-Ma2, anti-Ma1, anti-CV2
Extrapyramidal syndrome	SCLC, thymoma, testicular	anti-CV2, anti-Hu, anti-LGI1, anti-Ma2
Brainstem encephalitis	SCLC, breast, others	anti-Hu, anti-Ri, anti-GABABR, anti-IGLON5
	Testicular	anti-Ma2, anti-KLHL11
Optic neuritis	SCLC, lymphoma, others	anti-CV2, anti-AQP4
Retinal degeneration	SCLC, others	antirecoverin, others
	Melanoma	antibipolar cell
Myelopathy	SCLC, thymoma, breast, others	anti-CV2, anti-amphiphysin, anti-AQP4, anti-GlyR, anti-GFAP
Stiff-person syndrome	Breast, SCLC, thymoma, others	antiamphiphysin, anti-Ri, anti-GAD, anti-GlyR
Motor neuron disease	SCLC, others	anti-Hu, anti-Ri
Sensory neuronopathy	SCLC, others	anti-Hu, anti-CV2, ANNA-3, anti-Ma1, antiamphiphysin
	Plasma cell dyscrasias	antidisialosyl gangliosides
Neuromyotonia	Thymoma, SCLC, others	anti-Caspr2
Sensorimotor polyneuropathy	SCLC, others	anti-Hu, anti-CV2, ANNA-3
	Plasma cell dyscrasias	anti-MAG
Autonomic insufficiency	SCLC, thymoma	anti-Hu, anti-ganglionic AchR, anti-Caspr2
Lambert-Eaton syndrome	SCLC	anti-VGCC
Myasthenia gravis	Thymoma	anti-AchR, antistriated muscle, anti-MUSK

*This table lists only the most frequent associations. For each of the clinical syndromes, a number of other tumor types may be associated. A varying proportion of patients with each of the syndromes is "antibody-negative" or has 1 or more “atypical” autoantibody specificities that do not fit the well-characterized patterns listed.

**At least some antibodies previously thought to bind to voltage-gated potassium channels have been shown to bind to 1 or more proteins closely associated with potassium channels in the central and peripheral nervous systems. These proteins include leucine-rich, glioma-inactivated protein 1 (LGI1) and contractin-associated protein-2 (Caspr2) (21).

Abbreviations: AchR: acetylcholine receptor; AMPAR: amino-3-hydroxy-5-methylisoxazole-4-propionic acid receptor; AQP: aquaporin; Caspr2: contractin-associated protein-2; GABABR: GABAB receptor; GAD: glutamic acid decarboxylase; mGluR: metabotropic glutamate receptor; GlyR: glycine receptor; IGLON5: immunoglobulin-like cell adhesion molecule 5; KLH11: Kelch-like protein; LGI1: leucine-rich glioma-inactivated protein 1; MAG: myelin-associated glycoprotein; NMDAR: N-methyl-D-aspartate receptor; PKC: protein kinase-C; VGCC: voltage-gated calcium channel

Epidemiology

Many types of tumors have been associated with paraneoplastic disorders, but for most paraneoplastic neurologic syndromes, an overrepresentation of 1 or a few particular neoplasms is seen (12). Neuroblastoma and small-cell lung carcinoma are the tumors most often associated with paraneoplastic phenomena in children and adults, respectively. Paraneoplastic opsoclonus-myoclonus occurs in 2% to 3% of children with neuroblastoma. Paraneoplastic disorders occurring in association with other childhood tumors are rare. Approximately 1% to 3% of patients with small cell lung carcinoma develop Lambert-Eaton myasthenic syndrome or another paraneoplastic syndrome (35). Other tumors overrepresented among patients with paraneoplastic syndromes include breast carcinoma, ovarian carcinoma, ovarian teratoma, Hodgkin lymphoma, thymoma, and testicular germ cell tumors.

Prevention

The only preventive measure relates to the strong connection between cigarette smoking and small cell lung carcinoma; abstinence from smoking cigarettes may help prevent small cell lung carcinoma.

Differential diagnosis

Differential diagnosis in a patient with a suspected paraneoplastic disorder varies, depending on whether cancer is a known diagnosis. Among patients with neurologic dysfunction and a known cancer diagnosis, the level of suspicion for a paraneoplastic disorder depends on the neurologic syndrome, tumor histology, and presence of antineuronal antibodies. For example, small cell lung carcinoma or Hodgkin lymphoma are much more often associated with paraneoplastic disorders than squamous cell lung carcinoma or non-Hodgkin lymphoma. Tumor metastases and neurotoxicity of cancer treatments are far more common than paraneoplastic disorders and should always be considered, as should metabolic derangements and CNS infection.

For patients without a previous cancer diagnosis, the level of suspicion for a paraneoplastic disorder depends on patient age, gender, risk factors (especially cigarette smoking), the neurologic syndrome, and the presence of antineuronal antibodies. Several syndromes should always raise the possibility of a paraneoplastic etiology, including Lambert-Eaton myasthenic syndrome, subacute cerebellar degeneration, severe sensory neuronopathy, limbic encephalopathy, and opsoclonus-myoclonus (13; 12). For each of these syndromes, a rather short list exists of possible associated neoplasms.

There is no clinical neurologic syndrome that is absolutely associated with neoplasia; every neurologic "paraneoplastic syndrome" can occur in patients without a tumor, with the proportion of paraneoplastic versus nonparaneoplastic etiologies varying among individual syndromes. With increasing recognition of patients with autoantibodies such as anti-NMDAR antibodies, anti-GABAB receptor antibodies, and antibodies against the VGKC complex, CNS paraneoplastic disorders can be viewed as a subset of autoimmune encephalitides in adults and children (16; 19; 22).

Diagnostic workup

Brain MRI scans in patients with paraneoplastic encephalomyelitis may (or may not) show focal lesions in the cerebral cortex, limbic system, basal ganglia, brainstem, or spinal cord in patients with corresponding clinical involvement. In the acute phase of paraneoplastic cerebellar degeneration, opsoclonus-myoclonus, or brainstem encephalitis, brain MRI scans are usually normal but may show lesions in the cerebellum or brainstem. Brain MRI scans in most patients with paraneoplastic limbic encephalitis show areas of abnormal T2-weighted or FLAIR signal in the mesial temporal lobe and amygdala bilaterally and, less commonly, in the hypothalamus and basal frontal cortex. Most patients with paraneoplastic myelopathy have abnormal spine MRI scans (10). If present, MRI lesions in patients with CNS paraneoplastic disorders are nonspecific, and one cannot distinguish a paraneoplastic etiology from other conditions.

Most patients with a CNS paraneoplastic syndrome have abnormal cerebrospinal fluid, including some combination of mildly elevated protein, mild mononuclear pleocytosis, elevated IgG index, or oligoclonal bands. Normal cerebrospinal fluid does not exclude a paraneoplastic diagnosis. In addition, when normal initially, one can consider repeating cerebrospinal fluid testing within a week or 2, which in some instances may finally show evidence of immunologic response intrathecally.

One must note that serum and cerebrospinal fluid have different sensitivities for the antibodies of interest, and sending out both serum and cerebrospinal fluid paraneoplastic panel or autoimmune encephalitis panel has become the standard of care.

In a retrospective study of confirmed autoimmune encephalitis cases, FDG-PET CT scan of the brain was shown to be 1 of the earliest abnormalities detected, even when MRI/EEG/cerebrospinal fluid testing were unremarkable. The most common finding was hypometabolism of various brain regions (33).

In this author’s opinion, one must exercise caution and never rely on an abnormal FDG-PET CT scan of the brain as a sole indicator of autoimmune encephalitis or a reason to administer an immunologic agent, as FDG-PET scans are frequently abnormal in patients with other disorders. To date, no specific pattern of FDG-PET scan abnormality has been shown to be specific for autoimmune encephalitis.

Nerve conduction studies and electromyography are valuable in characterizing the clinical syndrome in patients with suspected peripheral nervous system paraneoplastic disorders (eg, Lambert-Eaton myasthenic syndrome, neuromyotonia, or sensory neuronopathy), but cannot in themselves differentiate a paraneoplastic versus nonparaneoplastic etiology.

The tumor workup for adults should include CT or MRI scans of the chest and abdomen (37). Chest CT or MRI scanning is clearly more sensitive than "plain" chest x-rays in detecting a lung neoplasm. Women should additionally have mammography and examination and imaging of the pelvic organs. Testicular ultrasound in young men may detect a small germ cell tumor or microcalcifications indicative of microscopic intratubular germ cell tumor (26). Exploratory laparoscopy or blind oophorectomy have enabled discovery of small ovarian teratomas in some patients with anti-NMDAR-associated encephalitis (05). Total-body fluorodeoxyglucose positron emission tomography (FGD-PET) with fused CT scanning of the chest and abdomen may demonstrate a neoplasm in patients with suspected paraneoplastic disorders (with or without autoantibodies) in whom other imaging studies are negative or equivocal (17; 28; 27). For patients with any of the syndromes, it is not uncommon for the tumor to be found only after repeated searches. As paraneoplastic syndromes can precede the tumor detection by at least several years, when the initial tumor workup is negative, it should be repeated yearly for the first 5 years, though no specific guideline has been established to date.

Good but not perfect correlations exist among individual paraneoplastic syndromes, antineuronal antibody specificities, and associated tumor types (Table 2). The practical clinical value of antineuronal antibodies is that, when present, they greatly increase the index of suspicion for a paraneoplastic condition, and the type of antibody can help guide the search for the associated tumor. Antineuronal antibody assays do, however, have important practical clinical limitations:

	(1) Heterogeneity exists, in that a given clinical syndrome (eg, cerebellar degeneration) may be associated with 1 of several autoantibodies.
	(2) Conversely, a given autoantibody (eg, anti-Hu) may be associated with a variety of clinical presentations.
	(3) For most, if not all, neurologic syndromes, some patients have antineuronal autoantibodies and yet never develop a demonstrable tumor. The prime examples are Lambert-Eaton syndrome, neuromyotonia, and limbic encephalitis. Presence of antibodies, therefore, does not absolutely indicate an underlying neoplasm.
	(4) Some autoantibodies are present at low titers in tumor patients without any accompanying clinical neurologic manifestations.
	(5) Patients with a suspected paraneoplastic syndrome may not have demonstrable antineuronal antibodies or may have "atypical" or incompletely characterized antibodies not detected in commercially available assays. Some patients initially determined to be "negative" for the more common antibodies turn out eventually to have newly recognized antibodies as the list of paraneoplastic antibodies grows (23; 06; 22). A negative antibody assay, therefore, does not rule out the possibility of a paraneoplastic disorder and the presence of an underlying neoplasm.

The ever-growing list of antineuronal antibodies associated with paraneoplastic disorders, as well as the heterogeneity of clinical-antibody associations, complicate decision-making for the clinician in deciding "which antibodies to order." Several of the antibodies have been described in only a small number of patients, and the practical utility of ordering all antibodies in all patients is highly debatable. Many laboratories perform a panel of different antibody assays (depending on the clinical syndrome) or perform a screening with immunohistochemistry or immunoblotting to be followed by specific antibody identification if necessary.

Management

Management of patients with known or suspected paraneoplastic syndromes includes the following components:

(1) verification that the disorder is, in fact, paraneoplastic; (2) diagnosis of the associated tumor; (3) treatment of the tumor; (4) suppression of the autoimmune effectors causing neuronal injury.

Several neurologic syndromes are considered "classical" paraneoplastic syndromes in that they are strongly associated with cancer (13). These include multifocal encephalomyelitis, limbic encephalitis, subacute cerebellar degeneration, opsoclonus-myoclonus, subacute sensory neuronopathy, chronic gastrointestinal pseudo-obstruction, Lambert-Eaton myasthenic syndrome, and dermatomyositis. Patients presenting with any of these syndromes should be thoroughly investigated for an associated neoplasm, regardless of the presence or absence of onconeural antibodies.

The workup to find an associated tumor should be guided by the known associations between particular clinical syndromes and tumor types and by the type of autoantibodies, if present.

When the tumor is definitely diagnosed, it should be treated with the appropriate surgical, chemotherapeutic, or radiation measures. For at least some patients with CNS paraneoplastic disorders, successful tumor treatment is associated with better neurologic outcome (38). Patients with some syndromes, most notably limbic encephalitis and opsoclonus-myoclonus in adults or children, often show neurologic improvement solely with successful tumor therapy. Some patients with cerebellar degeneration and Hodgkin lymphoma, and exceptional patients with carcinoma, improve neurologically after successful tumor treatment. Some patients with peripheral nervous system paraneoplastic disorders also show neurologic improvement with successful tumor treatment. This includes Lambert-Eaton syndrome, neuromyotonia, and probably vasculitic neuropathy and polymyositis as well.

If paraneoplastic disorders are truly autoimmune diseases, they should theoretically respond to immunosuppressive or immunomodulatory treatment. Several factors make it difficult to interpret the published literature regarding immunotherapy for paraneoplastic disorders:

	• These syndromes are relatively rare. For some of the syndromes, only a handful of well-characterized published cases can be found.
	• Most reports are anecdotal, and nearly all published series are retrospective.
	• A reporting bias exists, in that patients who respond to treatment are more likely to be published than those who do not respond.
	• For some syndromes, pharmacologic treatments can improve neurologic symptoms independent of tumor treatment or immunotherapy; examples include Lambert-Eaton syndrome (pyridostigmine, 3,4-diaminopyridine, or guanidine), limbic encephalitis (psychoactive and antiepileptic drugs), neuromyotonia (phenytoin and carbamazepine), stiff-person syndrome (diazepam and baclofen), and opsoclonus-myoclonus (clonazepam or valproate).
	• Paraneoplastic cerebellar degeneration, sensory neuronopathy or encephalomyelitis, and other syndromes often stabilize spontaneously (although at a level of severe disability) so that it is difficult to interpret reports of "disease stabilization" with immunotherapy.
	• Perhaps the greatest difficulty in evaluating immunotherapy for paraneoplastic disorders is the confounding effect of tumor treatment. When patients receive concomitant tumor treatment and immunotherapy, it becomes difficult to discern the impact of each therapy on the neurologic outcome. For many syndromes, increasing evidence shows that immunotherapy is more likely to be effective when the tumor is also treated successfully.

Factors that interact in influencing the response to immunotherapy include the neuroanatomic site (central vs. peripheral), the cellular location of the onconeural target antigens (neuronal cell surface vs. intracellular), and the proven or presumed mechanisms of neuronal injury (antibody-mediated vs. cell-mediated). In general, syndromes affecting the peripheral nervous system are more likely to improve with tumor treatment or immunosuppressive treatment than are CNS syndromes. Syndromes caused by autoantibodies reacting with neuronal cell surface receptors or ion channels are more likely to respond to immunotherapy, probably because the antibodies do not usually cause axonal degeneration or neuronal cell death. Prime examples of this are the Lambert-Eaton myasthenic syndrome and neuromyotonia. Patients with encephalitis associated with anti-NMDAR antibodies, anti-GABABR antibodies, or antibodies against the voltage-gated potassium channel complex (anti LGI1 or anti-Caspr2) generally have a good neurologic outcome after immunotherapy (23; 19; 06; 22). Sensory neuronopathy and autonomic insufficiency involve the peripheral nervous system, but patients generally do not respond to therapy, which is probably because neuronal cell bodies in the dorsal root ganglia and autonomic ganglia are irreversibly injured or killed by the autoimmune response.

Unfortunately, the 2 most prevalent paraneoplastic CNS syndromes in adults, ie, encephalomyelitis or sensory neuronopathy associated with anti-Hu antibodies and cerebellar degeneration associated with anti-Yo antibodies, usually have a poor neurologic prognosis despite aggressive tumor treatment and a variety of immunosuppressive therapies (38). Patients with other CNS syndromes, including opsoclonus-myoclonus, limbic encephalitis, and stiff-person syndrome, have a somewhat higher likelihood of neurologic improvement, suggesting that the immune-mediated neuronal dysfunction or injury is less severe or of a sort more likely to be reversible. Even for the "unfavorable" syndromes, such as encephalomyelitis and cerebellar degeneration, a few patients do show a meaningful neurologic response to immunotherapy. For these few responders, the only factors that sometimes correlate with neurologic improvement are successful tumor treatment and the duration and severity of neurologic deficits prior to diagnosis and initiation of therapy. For patients who have already stabilized at a plateau of severe neurologic disability for more than several weeks, subsequent improvement with any intervention is not impossible but extremely unlikely. Therefore, the decision whether to try immunosuppressive therapies must be based on the particular syndrome and on the individual patient's circumstances.

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, cautious 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. Plasma exchange could arguably be reserved for syndromes that are definitely antibody-mediated, but reports have been made of improvement in patients with other syndromes as well (39).

Some paraneoplastic disorders may respond better to repeated plasma exchanges, whereas others may have a better record at responding to IVIg. In addition, when the antibody specificity is not yet available, many clinicians will choose to proceed with 5 plasma exchange sessions approximately every other day for a rapid antibody removal and then administer intravenous immunoglobulins for a more extended immune modulation. Chronic oral or pulsed intravenous cyclophosphamide has also been used for a variety of paraneoplastic disorders, with no definite evidence establishing an optimal schedule or dose. Azathioprine or cyclosporine has been used with benefit in some patients with Lambert-Eaton syndrome, with little published information regarding their efficacy in other paraneoplastic disorders. The monoclonal antibody rituximab, which depletes B lymphocytes, produced neurologic improvement in children with neuroblastoma and opsoclonus-myoclonus and in young people with anti-NMDAR encephalitis (32; 04). Rituximab and/or cyclophosphamide have been used as second-line therapy for adult patients with multifocal or limbic encephalitis who do not respond well to corticosteroids, IVIg, or plasma exchange (23). There are studies of tacrolimus plus prednisone or of sirolimus, which are agents that specifically inhibit activated T-lymphocytes in patients with anti-Hu or anti-Yo-associated syndromes (31; 08).

Several potential explanations exist for the disappointingly poor response to immunotherapy in many patients. The continuing presence of even a small tumor burden seems to provide an "antigenic drive" for further neuronal injury. It is also likely that current immunotherapies do not adequately gain access to the central nervous system and do not effectively abrogate an ongoing autoimmune response that is "sequestered" in the central nervous system. Unfortunately, for many central syndromes, it is likely that patients have already suffered neuronal death or irreversible injury by the time the diagnosis of a paraneoplastic disorder is made.

Theoretical concern exists that if paraneoplastic disorders arise 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, no definite evidence shows that patients given immunosuppressive treatment have a worse tumor outcome.

Special considerations

Pregnancy

When a paraneoplastic disorder arises during pregnancy, the search for a tumor, tumor removal, and adequate immunologic therapy are vitally important for the maternal-fetal outcomes. The side effects of diagnostic imaging (for example, CT chest/abdomen/pelvis) need to be weighed against the risks associated with the lack of a timely diagnosis and treatment.

Most pregnant women are able to safely undergo cerebrospinal fluid testing and can have MRI/ultrasound modalities employed as an initial strategy of tumor detection.

Immunologic therapies such as intravenous steroids and intravenous immunoglobulins can generally be safely administered in pregnant women (with specific trimester precautions) when benefits clearly outweigh the risks. This decision should be always discussed with an obstetrician in charge of the patient’s care.

In patients not responding to the first-line treatment who are rapidly deteriorating (for example, seizing), a second-line therapy such as rituximab should be instituted. A limited amount of literature and experience suggests that rituximab can be safely administered to pregnant women in life-threatening situations and in disorders that carry a high risk of morbidity and mortality to the mother and fetus if not treated on a timely manner. In the world of neuroimmunology, patients with neuromyelitis optica frequently receive rituximab during pregnancy. A retrospective study, though based on a small sample size (102), showed no increased risk of spontaneous abortion or an increased risk of fetal malformation in babies exposed to rituximab in utero (07). Long-term effects of in utero exposure to rituximab have never been studied in an organized fashion.

Anesthesia

Patients with Lambert-Eaton myasthenic syndrome may develop severe, prolonged weakness, respiratory insufficiency, or both with neuromuscular blocking agents.