Paraneoplastic Neurological Disorders of the Central Nervous System

Many of these disorders occur in association with immune responses against intraneuronal antigens expressed by the underlying cancer (paraneoplastic or onconeuronal antigens). These immune responses are characterized by the presence of antibodies and cytotoxic T cell responses against the onconeuronal antigens. As far as the antibodies are concerned, there have been several attempts to reproduce PND by passive transfer of serum or immunoglobulin G (IgG) to animals or by immunization with recombinant antigens that prime B cell responses.^4–6 These procedures did not result in neurological dysfunction, although immunization did result in high titers of serum antibodies.⁷ It has been argued that the antibodies cannot reach the intracellular antigens, but data from other immune-mediated disorders suggest that antibodies can enter cells and disrupt the function of the antigen.⁸ Prior studies may have failed in part because they did not reproduce the continuous intrathecal synthesis of antibodies that occurs in patients with PND.

As far as the T cell response is concerned, a pathogenic role was initially suggested by the neuropathological findings in autopsy studies of patients with PND.⁹ These studies demonstrated prominent infiltrates of inflammatory cells, later characterized as CD4⁺ and CD8⁺ T cells that are usually accompanied by microglial activation and gliosis (Fig. 101-1).^10,11 Analysis of the presence of antigen-specific T cells in the peripheral blood and brain infiltrates of patients and studies of antigen presentation by dendritic cells or fibroblasts modified to express the paraneoplastic antigens provide strong evidence of a role of cytotoxic T cells.^12–15 An attempt to model the disease in animals by priming the T cell response resulted in minimal perivascular inflammatory infiltrates but did not reproduce the disease or result in interstitial T cell infiltrates with neuronophagia that occur in patients.¹⁶

Figure 101-1 Activated cytotoxic T cells (dark brown staining) in the brain of a patient who died of anti-Hu–associated paraneoplastic encephalomyelitis. The T cells have been immunolabeled with TIA, a marker of activated cytotoxic T cells. Small arrows indicate T cells; arrowhead indicates a neuron undergoing degeneration.

Overall, these studies suggest that both antibody and cytotoxic T cell responses are probably necessary to cause neuronal injury. This concept is important not only for modeling the disease but also for the development of treatment strategies for PNDs of the central nervous system (CNS).

Paraneoplastic Neurological Disorders of the Peripheral Nervous System

For many PNDs of the peripheral nerve and muscle, the evidence of immune-mediated mechanisms is also abundant, but the target antigens are less well defined than in PNDs of the CNS. Evidence of immune mechanisms is found mainly by the detection of inflammatory infiltrates composed of T cells.

A direct pathogenic role of antibodies has been demonstrated in only a few disorders that can occur with or without a cancer association, including (1) neuromyotonia and antibodies to voltage-gated potassium channels, (2) a subset of autonomic neuropathies and antibodies to the ganglionic acetylcholine receptor (AChR), and (3) disorders of the neuromuscular junction, such as myasthenia gravis in association with antibodies to the AChR of the neuromuscular junction and the Lambert-Eaton myasthenic syndrome (LEMS) in association with antibodies to P/Q-type voltage-gated calcium channels. In contrast to PNDs of the CNS, in which the antigens are usually intraneuronal, the antigens of antibody-mediated PNDs are receptors or ion channels expressed on the surface of nerves, or at the preneuromuscular or postneuromuscular synapse.

Effects of Paraneoplastic Immunity on the Tumor

It has been suggested that paraneoplastic immunity is clinically effective in controlling tumor growth, accounting for the small size and limited metastatic burden of tumors associated with PNDs. However, despite the demonstration that some patients with PNDs develop cytotoxic T cell responses to tumor antigens, the oncological outcomes in studies of hundreds of patients with antibody-associated PND do not significantly differ from those of patients without PNDs.^17–23 In addition, investigators who examined whether immunotherapy favored tumor growth did not identify a change in tumor behavior.^24,25 By and large, these studies suggest either that the efficacy of the antitumor immune response is minimal or that the cancer usually overcomes the antitumor immune effect.¹

Recognition of the Neurological Syndrome

An extensive group of disorders similar to PNDs may occur in the absence of cancer (Table 101-3). However, some syndromes are associated with cancer much more frequently than others, or the clinical features are characteristic enough that they readily suggest a paraneoplastic etiology. These syndromes are considered “classical PNDs” (see Table 101-1). Other syndromes may result from paraneoplastic mechanisms but occur more frequently in the absence of cancer. These syndromes are considered “nonclassical” and necessitate a more extensive differential diagnosis. For example, a brainstem syndrome, chorea, or the Guillain-Barré syndrome may be paraneoplastic manifestations of cancer but are mostly not cancer related, or their development in a patient with cancer may simply be coincidental.

TABLE 101-3 Differential Diagnosis of Paraneoplastic Neurological Disorders of the Central Nervous System

Ara-C, cytosine arabinoside; 5-FU, 5-fluorouracil; GAD, glutamic-acid decarboxylase; HHV, human herpesvirus; HSV, herpes simplex virus; VGKC, voltage-gated potassium channel.

Most PNDs develop and progress rapidly until stabilization in a few weeks or months, causing severe disability. Patients often become wheelchair bound or bedridden over a short period of time. PNDs that affect the CNS, dorsal root ganglia, or proximal nerve roots are often accompanied by lymphocytic pleocytosis, an elevated IgG index, the presence of oligoclonal bands, or intrathecal synthesis of paraneoplastic antibodies. However, similar CSF abnormalities can be encountered in any inflammatory or immune-mediated disorder of the CNS, and some patients with PNDs may have normal findings on CSF studies. In most instances, CSF studies are necessary to rule out other cancer complications, such as leptomeningeal metastasis.²⁶

All patients with PNDs of the CNS and some peripheral nerve syndromes (e.g., plexopathies) should undergo neuroimaging evaluation of the involved area. Magnetic resonance imaging (MRI) is the best technique for ruling out metastatic lesions or other complications that may suggest a PND. In most PNDs of the CNS, the function of the blood-brain barrier is preserved, and therefore the affected brain regions are rarely enhanced with contrast material. The abnormalities are usually demonstrated with T2-weighted imaging and fluid-attenuated inversion recovery (FLAIR) sequences. In syndromes such as limbic encephalitis with predominant hippocampal involvement (short-term memory loss, seizures), the MRI findings are often suggestive of the syndrome, although the etiology could be nonparaneoplastic.²⁷ There is also increasing evidence that brain [F18] fluorodeoxyglucose positron emission tomography (FDG-PET) has diagnostic usefulness in PNDs.²⁸ In the early stages of PNDs, FDG-PET may show hypermetabolism in the abnormal brain regions identified by MRI, but in some patients, the MRI findings are normal.²⁹

Biopsy of an abnormal brain region identified by MRI or FDG-PET may be considered if a neoplastic process is suspected or if the clinical, CSF, and MRI findings are unusual. Abnormalities supporting but not specific to PND include infiltrates of mononuclear cells, neuronophagic nodules, neuronal degeneration, microglial proliferation, and gliosis.³⁰

Demonstration of Associated Cancer

PNDs usually develop at early stages of cancer, and therefore the tumor (or tumor recurrence) may be difficult to demonstrate. Modern imaging techniques are able to demonstrate small tumors that were often missed by techniques used previously. In most instances, the tumor is revealed by computed tomography of the chest, abdomen, and pelvis. The type of syndrome and paraneoplastic antibody may suggest a specific underlying tumor and the need for additional tests, such as mammography or ultrasonography of the pelvis or testes. Whole-body FDG-PET is very useful in demonstrating occult neoplasms or small metastatic lesions that may be more accessible for biopsy than the primary tumor is (Fig. 101-2).³¹ In addition to imaging studies, serum cancer markers such as carcinoembryonic antigen, CA-125, CA-15.3, or prostate-specific antigen are helpful. All patients with a neuropathy of unclear etiology should be examined for the presence of a monoclonal gammopathy in the serum and urine and, if results are positive, undergo a skeletal survey and bone marrow biopsy; these studies may uncover a malignant plasma cell dyscrasia, amyloidosis, or B cell lymphoma.²

Figure 101-2 A, Body fluorodeoxyglucose positron emission tomography (FDG-PET) demonstrates the tumor in a patient with anti-Yo antibodies. Abnormal FDG uptake demonstrates right axillary adenopathy, revealed by biopsy to be an adenocarcinoma. The detection of serum and cerebrospinal fluid anti-Yo antibodies indicated that the patient had paraneoplastic cerebellar degeneration and that the primary tumor was likely to be in the breast. B, A section of rat cerebellum has been incubated with the serum of the patient; note that the anti-Yo antibody reacts predominantly with the cytoplasm of Purkinje cells and, to a lesser degree, with the cytoplasm of cells of the molecular layer.

Close oncological surveillance should be undertaken in patients with classical PNDs with or without paraneoplastic antibodies and in patients with nonclassical PND and paraneoplastic antibodies. It is recommended that patients undergo periodic cancer screening for at least 5 years after diagnosis of PNDs; in 90% of patients, the underlying tumor is discovered within the first year of PND symptom manifestation. Patients whose cancer is in remission and who develop PNDs should be examined for tumor recurrence.

Detection of Paraneoplastic Antibodies

Paraneoplastic antibodies are antibodies whose presence serves as a marker of the paraneoplastic origin of a neurological syndrome. Several concepts are important in testing for paraneoplastic antibodies (Table 101-4). First, antibodies are present in approximately 60% of patients with PNDs of the CNS; therefore, the absence of antibodies does not preclude a paraneoplastic syndrome. Second, paraneoplastic antibodies may be identified (usually at low titers) in the serum of a variable proportion of patients with cancer but without PND (i.e., anti-Hu and anti-CV2/CRMP5 in 20% and 10% of patients with SCLC, respectively).^32,33 Third, in PNDs of the CNS, the antibodies are found in serum and CSF; detection of CSF antibodies is a strong indicator that the associated neurological syndrome is paraneoplastic. Fourth, most PNDs of the peripheral nerve or muscle are not associated with paraneoplastic antibodies, except for anti-Hu (Fig. 101-3) and anti-CV2/CRMP5 antibodies. Fifth, not all paraneoplastic antibodies have the same sensitivity and specificity; on the basis of their clinical relevance, the paraneoplastic antibodies are classified in two categories: well-characterized paraneoplastic antibodies and partially characterized antibodies (see Table 101-4).³

TABLE 101-4 Paraneoplastic Antibodies

PCD, paraneoplastic cerebellar degeneration; PSN, paraneoplastic sensory neuronopathy.

* Some patients harbor Ma1 and Ma2 antibodies; the presence of Ma1 is usually associated with predominant brainstem and cerebellar involvement and with tumors other than testicular neoplasms. The prognosis in patients with tumors other than testicular neoplasms is poorer than that of patients with Ma2 antibodies and testicular neoplasms.

Figure 101-3 Expression of Hu proteins by human dorsal root ganglia neurons. Anti-Hu immunolabeling of a section of human dorsal root ganglia obtained from the autopsy study of a neurologically normal individual. Note that the Hu proteins are expressed specifically by the nuclei and, to a lesser degree, by the cytoplasm of neurons (brown staining). The surrounding satellite cells do not contain Hu proteins.

Well-characterized paraneoplastic antibodies include anti-Hu, Yo, Ma2, Ri, CV2/CRMP5, and amphiphysin. These six antibodies and the corresponding antigens have been characterized by different laboratories and reported in large series of patients with PNDs. Detection of any of these antibodies strongly supports the diagnosis of PND even if no tumor is found at initial evaluation. Some antibodies are more syndrome specific than others; for example, anti-Yo antibodies are almost always accompanied by cerebellar degeneration, and anti-Ma2 antibodies are almost always accompanied by limbic or upper brainstem dysfunction, whereas anti-Hu or anti-CV2/CRMP5 antibodies are accompanied by a much wider spectrum of symptoms.

Partially characterized antibodies are those with which clinical experience is limited or for which the target antigens are unknown. Until there is more experience, detection of any of these antibodies is of limited diagnostic value, and the management of affected patients should be similar to that of patients without paraneoplastic antibodies, including extensive clinical, CSF, and neuroimaging evaluations to rule out other, more frequent complications of cancer.

Several antibodies, including P/Q-type voltage-gated calcium channels, voltage-gated potassium channels, and nicotinic or ganglionic AChR antibodies can be detected in both the paraneoplastic and nonparaneoplastic forms of the associated disorder (Table 101-5