U. Wurster
The generally used description of these antibodies as antineuronal antibodies (ANab) is historical and inaccurate. It is used to refer collectively to all types of antibodies against every possible component or structure of the peripheral and central nervous systems. Since these antibodies are not limited to neurons, the term antineural antibody would be more appropriate, but so far it is has not come into common use.
Diseases with Suspected Involvement of ANab
Although a direct pathological effect of ANab has been demonstrated only for myasthenia gravis (acetylcholine receptor blockade), Lambert-Eaton syndrome (calcium channel blockade), and the retinopathy caused by the binding of recoverin, the detection of ANab is often of great diagnostic value. Currently, it must be said, the only ANab that are of established diagnostic importance are those that have been described for a large enough number of patients (> 10) and for which standardized, reproducible tests are available. The variously suspected involvement of ANab in Parkinson's disease, AIDS dementia, and particularly schizophrenia (Giovannoni and Dale, 2001) is certainly very interesting, but is so far without any practical importance.
Gluten Ataxia
Hadjivassiliou et al. (2003 a) put forward the controversial thesis that antibodies against gliadin (a gluten contained in wheat) are responsible not only for intestinal symptoms (celiac disease, sprue), but may in addition—or even exclusively—affect the cerebellum by cross-reacting with Purkinje cells. A gluten-free diet has proven successful in a study involving 43 patients (Hadjivassiliou et al., 2003 a). Even though the prevalence of gliadin antibodies reported by Hadjivassiliou et al. was clearly higher in sporadic ataxia (41%) than in hereditary forms of ataxia (14%) and in normal persons (12%), the test for gliadin antibodies is not specific enough for gluten ataxia.
Hence, the direct detection of gliadin antibody reactivity on cerebellar sections would be of great importance. However, the antibody titers are relatively low (a maximum of 1:800), and the diffuse Yo-like staining pattern of the cytoplasm of Purkinje cells is evidently insufficient for reliable evaluation (Hadjivassiliou et al., 2003 b). The intended characterization of antigens by immunoblotting is still pending.
Post-streptococcal Autoimmune Diseases of the CNS
Sydenham's chorea, seen mostly in children in developing countries and occurring 3–5 months after infection with group A β-hemolytic streptococci, is regarded as a model of autoimmune diseases of the CNS that are induced by molecular mimicry.
Etiology. A cross-reaction between antibodies against M proteins of certain streptococcal strains (M5, M6, M19, and M24) and as yet incompletely characterized antigens in the basal ganglia has been postulated as a mechanism. Sections of human basal ganglia yielded selective staining of neuronal tracts, whereas Western blotting of homogenates from basal ganglia showed immune reactivity directed mainly against antigens with molecular weights of 40, 45, and 60 kDa.
ABGA. The assumption of the existence of anti-basal ganglia antibodies (ABGA) directed to an autoantigen seems justified, since they were detected in 100% of patients with the acute form of Sydenham's chorea and in 69% of patients with the persistent form, but in no healthy control persons. Furthermore, the antibody reactivity is restricted to basal ganglia, and is absent in the cerebellum, cortical neurons, and myelin (Church et al., 2002).
PANDAS. The hypothesis that autoimmune reactions induced by streptococci might not be responsible for Sydenham's chorea alone was first formulated more than 10 years ago (Budman et al., 1997) in connection with a group of children with compulsive-obsessive disorder or tic disorders occurring in association with streptococcal infections (p ediatric a utoimmune n europsychiatric d isorders a ssociated with s treptococcal infections, PANDAS) (Snider and Swedo, 2003). The concept has since been expanded to include a whole spectrum of neuropsychiatric diseases, particularly extrapyramidal movement disorders (chorea, tics, dystonia) and psychiatric manifestations (anxiety and compulsive disorders, attention deficit hyperactivity disorder) (Dale, 2003). For some of these patients (e. g., about 20% of the patients with Tourette's syndrome), evidence of ABGA means that an autoimmune reaction of this kind is at least conceivable. However, to date most reports on ABGA have originated from the same research group in London (Dale, 2003) and have not yet been independently confirmed.
The detection of ABGA by immunofluorescence using a 1:10 dilution of the serum is not very sensitive, and a common antigen is not detected by Western blot. Should it become possible to identify one or more specific antigens, this would open up important perspectives for the diagnosis and treatment of a relatively common disorder.
Diseases with Confirmed Involvement of ANab
Autoantibodies to Glutamate Decarboxylase
Physiology. The enzyme glutamate decarboxylase (GAD) catalyzes the production of the neurotransmitter GABA from glutamate. Two isoforms (65 kDa and 67 kDa) occur at higher concentration in neurons and in the islet cells of the pancreas.
Associated diseases. Antibodies against GAD are present at low concentration in diabetes mellitus type I; they are mainly directed against conformational epitopes. A titer 10–1000 times higher with reactivity against linear epitopes of both isoforms is found in about 70% of patients with stiff person syndrome. A pronounced increase in GAD antibodies was also described in strictly cerebellar chronic ataxia with late manifestation of insulin-dependent diabetes, medically refractory epilepsy, and autoimmune polyendocrinopathy.
Histochemical detection. Because of the very high antibody titers in stiff person syndrome, it is usually possible to detect GAD antibodies by histochemistry. Sections of monkey cerebellum reveal a spotted “leopardskin” staining pattern of the neuropil of the granular layer. Evidently, after perfusion fixation there is less diffusion of the antigen, so the spotted pattern of GABAergic structures in the molecular and granular layers and in the baskets around the Purkinje cells is preserved. The reactivity of GAD antibodies with beta-islet cells in pancreas sections can be used to back up the results of the staining.
Confirmation. A dot blot using GAD65/67 (Medipan Diagnostica, Selchow, Germany) and a line blot using GAD65 (Imtec Immundiagnostika, Berlin, Germany) are available as confirmation tests. However, accurate quantification requires measurement by RIA. Since individual laboratories use very different methods, a reference range cannot be given. It should be ensured that the cut-off selected is high enough to exclude patients with diabetes (Murinson et al., 2004). To distinguish between neurological diseases and diabetes, the frequent occurrence of oligoclonal bands (> 60%) and the pronounced intrathecal synthesis of GAD antibodies in the CSF has proved helpful.
Paraneoplastic Antibodies
Definition of Paraneoplastic Neurological Syndrome
Paraneoplastic neurological syndromes are neurological diseases triggered by tumors. The cause of such a disease is not the tumor itself (i. e., it is not a result of metastasis, compression, radiation, neurotoxic chemotherapy, metabolic encephalopathy, or opportunistic infection) but rather an immunologically mediated distant effect due to cross-reactivity with the nervous system (see Chap. 10, “Paraneoplastic Neurological Syndromes”).
Paraneoplasia Pitfalls
Paraneoplastic syndromes are rare (< 1% of all tumor diseases) and thus may be easily overlooked in everyday clinical routine. They may affect any level of the nervous system and should therefore always be included in the differential diagnosis. Persons aged over 50 years, particularly smokers, are the main risk group. In two-thirds of cases, no tumor is known at the time of onset of neurological symptoms. Paraneoplasia may also be accompanied by atypical (psychiatric) symptoms and may occasionally have a fluctuating course or, rarely, a benign course (Chap. 10, “Paraneoplastic Neurological Syndromes”; Kaiser, 1999; Voltz, 2002; Darnell and Posner, 2003).
Occurrence. Paraneoplastic antibodies are of high diagnostic value: they can explain complex neurological symptoms, and they also allow a targeted search for particular associated tumors (Table 7.2).
Prevalence. The prevalence of individual paraneoplastic antibodies varies considerably (Pittock et al., 2003):
• Anti-Hu (0.72%), the most common antibody.
• Anti-Yo (0.52%), only in women.
• Anti-CRMP-5/CV2 (0.21%).
• Anti-Ri (0.04%).
• ANNA-3 (11/68 000 = 0.016%).
• Newly defined specificities include anti-SOX1 (previously called AGNA) and anti-NMDA receptor antibodies.
Immunohistochemistry. In Table 7.2, paraneoplastic antibodies are grouped according to their appearance in cerebellar sections as either nuclear staining (anti-Hu, anti-Ri, anti-Ma) or cytoplasmic staining of Purkinje cells (anti-Yo, anti-PCA-2, anti-Tr). Staining of the neuropil has been observed with anti-amphiphysin and anti-CRMP-5/CV2, but also with nonparaneoplastic antibodies against GAD and GQ1b ganglioside.
Pitfalls with Paraneoplastic Antibodies
Some antibodies have a variety of names. Naming them after the patient was introduced by J. B. Posner (Darnell and Posner, 2003). Also in use is a generic nomenclature by V. A. Lennon (Lennon, 1994).
Detection
Pitfalls
Because of the wide variety of methods used in the different laboratories (immunohistochemistry, immunofluorescence, Western blot, ELISA), the titers obtained also vary considerably and thus cannot be directly compared. Voluntary quality control is provided by Instand e. V. (Düsseldorf, Germany) for anti-Hu, anti-Yo, anti-Ri, and anti-amphiphysin.
Immunohistochemistry
Substrate. Normally, sections of the cerebellum are used as a substrate for detecting antineural antibodies. Apart from the large Purkinje cells (up to 40 μm in diameter), gray and white matter are present in perfect anatomical arrangement (arbor vitae). If the clinical symptoms suggest a particular localization, however, it may be necessary to use brain tissue from the clinically affected area as it has a selectively higher concentration of the relevant antigens (e. g., using the putamen for detecting basal ganglia antibodies). The use of human brain tissue is problematic because of the difficulties of acquisition and the extended post-mortem times. For detection of the established antineural antibodies, frozen sections (5–10 μm thick) of rat brain or cerebellum or commercially available monkey sections are usually equally suitable.
Fixation. In theory, unfixed sections have the advantage that antigens are preserved in their native form, but in practice the antigens are harder to identify accurately because of their diffusion and their poor structural preservation. On the other hand, the fixation methods that are generally used (acetone, methanol, ethanol, formaldehyde) may induce different staining behaviors. Predominant staining of oligodendrocytes has been described for CV2 antibody after perfusion fixation with paraformaldehyde, but not for CRMP-5 antibody—regarded as identical to CV2—after immersion fixation with formalin (Table 7.2) (Honnorat et al., 2001).
Mosaic sections. For detecting systemic antibodies and for detecting the reaction to autonomic nerves that is important for distinguishing anti-Hu from anti-Ri and ANNA-3, some laboratories (e. g., that of V. A. Lennon) use mosaic sections coated with cerebellum, stomach, and kidney right from the start. Mosaic sections are commercially available (The Binding Site, Bio-Rad Laboratories, Euroimmun AG) from cerebellum, cerebrum, liver, intestine, and peripheral nerve, and—optionally—also HEp-2 cells or cerebellum/stomach, as well as cerebellum/pancreas for GAD antibodies. For reasons of cost a stepwise diagnosis is recommended: cerebellar sections are screened first, and other tissues are tested only if the results are inconclusive.
Fig. 7.4 a, b Hu antibody.
a Purkinje cells show nuclear staining, with exclusion of the nucleoli. All neurons in the molecular (Mol) and granular (Gran) layers show nuclear staining as well. × 400.
b At higher magnification, 10–20 fine dots (arrows) are clearly visible in the cytoplasm of Purkinje cells. It is possible that this pattern is caused by another, associated antibody. × 2000.
Fig. 7.5 Yo antibody. Purkinje cells show coarse-granular staining of the cytoplasm, with exclusion of the nuclei. Cells in the molecular layer (Mol) are stained as well. × 400.
Fig. 7.6 Tr antibody. Purkinje cells show granular staining of cytoplasm and some dendrites. Fine dots are seen in the neuropil of the molecular layer (Mol). × 400.
Fig. 7.7 Amphiphysin antibody. Diffuse or granular staining (synapses) of the molecular (Mol) and granular (Gran) layers. × 125.
Fig. 7.8 CV2/CRMP-5 antibody. Diffuse staining of the neuropil of molecular (Mol) and granular (Gran) layers. Fainter staining of oligodendrocytes in the medulla (Med). × 200.
Immunostaining. The secondary antibodies against human IgG used for the immunostaining are either labeled with fluorescein or coupled to horseradish peroxidase or alkaline phosphatase; these labels can be further enhanced by using biotin/streptavidin. IgA and IgM antineural antibodies do not seem to occur (Greenlee et al., 2001). Usually, the test liquid is serum (samples can be sent by mail); it is used at a starting dilution of 1:100. For specific antibodies (anti-amphiphysin, anti-Tr, ABGA), lower dilutions may be recommended, although this considerably increases the nonspecific background. Unlike serum, fresh CSF yields very low background staining.
Antibodies in the CSF and serum. Since intrathecal synthesis of antibodies is common in paraneoplastic diseases, the fraction of autoantibodies in the total IgG is increased, so that interfering overlay by antibodies of different specificities is less noticeable. By titrating serum and CSF, the antibody index (AI) for intrathecal synthesis is calculated according to the following formula:
AI values higher than 1.5 are considered elevated. If only the peripheral nervous system is affected (e. g., in pure sensory polyneuropathy with Hu antibodies), no oligoclonal bands are observed in CSF, and there is no intrathecal synthesis of anti-Hu. Extensive local production seems to occur with CRMP-5/CV2 antibodies; in 37% of cases, higher titers were found in CSF than in serum (Yu et al., 2001). In 2/28 cases, detection of Tr antibodies was only successful in CSF (Bernal et al., 2003), and in 2/10 patients with limbic encephalitis, Ta (Ma-2) antibodies were below the detection limit in serum, whereas they were increased by several orders of magnitude in CSF (Gultekin et al., 2000).
Pitfalls
Low anti-Hu titers of up to 1:400 are also found in small-cell lung carcinoma in the absence of neurological symptoms. This fact confirms that paraneoplastic antibodies are primarily induced by the antigens of a tumor; they are therefore regarded as tumor markers. Despite intensive searching, including whole-body FDG-PET (Rees et al., 2001), in up to 15% of cases—depending on the antibody used—no tumor is found. The number becomes smaller with long-term followup, since a tumor may be detected up to 7 years later. Sometimes the tumors, which are often very small, are only found during meticulous autopsy. Even so, there are still cases in which despite all efforts no tumor is ever found. Apparently, in these cases the biologically appropriate immune reaction of the body against the tumor has led to its regression.
In contrast to earlier reports (Moll et al., 1993), the use of recombinant Hu, Yo, Ri, and amphiphysin antigens in systemic autoimmune diseases (71 patients with Sjögren's syndrome, 102 with lupus erythematosus) uncovered the presence of anti-Hu only in one Sjögren patient and one lupus patient (Benyahia et al., 2003).
When determining titers over time, titer levels should be standardized to equal IgG concentrations, as concentrations can vary considerably after plasmapheresis or the administration of immunoglobulins or continuous infusions. Decrease of titers during chemotherapy has been described for anti-Hu and anti-Yo, at least in serum. Where there is complete tumor remission, the test for anti-Hu may become negative (Llado et al., 2004).
Very rarely, at the very beginning of acute paraneoplastic disease, detection of the corresponding antibody may be very weak (histochemistry negative, Western blot borderline) or negative. Repeat serological testing and follow-up lumbar punctures are therefore highly desirable.
If none of the known paraneoplastic antibodies, including those against K+ and Ca2+ channels, can be detected, one should consider the possibility of atypical, not yet classified antibodies—above all when oligoclonal bands are present in the CSF, particularly those of type 3. Here, testing CSF rather than serum often yields clearer results. However, several criteria must be met before a new paraneoplastic antibody can be accepted. The most important is simultaneous detection of the potential antigen in tumor and nervous tissues, but not in systemic tissues.
Immunoblot
Application. Since immunohistochemical tests are less expensive, they are commonly used as a screening method. If a neural antibody is suspected, it should be confirmed by immunoblot (Chap. 4, “Electrophoretic Methods with Immunodetection”). This is necessary because it is often not possible to classify staining patterns obtained exclusively with cerebellar sections (Table 7.2), and the patterns may also be mimicked by (unknown) systemic antibodies.
Immunoblot (Chap. 4, “Electrophoretic Methods with Immunodetection”): Special Features
Separation of neural proteins according to their molecular weight (SDS polyacrylamide gel electrophoresis) with subsequent electrophoretic transfer of proteins onto a nitrocellulose or PVDF membrane:
Sample material: aqueous extracts of Purkinje cells or neurons isolated from the cerebellum by means of density gradients, or total homogenates or their centrifuged supernatants (myelin removed) from the cerebellum or other regions of the brain.
When using total homogenates, it is necessary to slightly overload the gels.
No reduction of proteins with mercaptoethanol or dithiothreitol (which otherwise is usual).
Non-neural antibodies. When using brain tissue for immunoblots, one repeatedly observes staining of bands that cannot be assigned to any known specificity. In almost all cases these most likely represent non-neuronal (i. e., systemic or natural) antibodies rather than genuine atypical neural antibodies. To prevent detection of non-neural antibodies, the serum may be first absorbed with acetone powder extract from liver. Another possibility is to include parallel runs of extracts from systemic tissues or HEp-2 cells. Alternatively, since intrathecal synthesis of paraneoplastic antibodies regularly occurs, serum and CSF of the patient can be adjusted to an identical IgG concentration of 5 mg/L and then analyzed in parallel. By comparing the resulting bands, it is possible to draw conclusions about the origin of the antibodies (Fig. 7.9).
Recombinant antigens. By using recombinant onconeural (paraneoplastic) antigens, immunoblots can be performed in a simplified form as line blots. In the commercially available kits (Euroimmun AG, Milenia Biotec, Ravo Diagnostika), the recombinant proteins are streaked onto nitrocellulose strips which only need to be incubated with serum or CSF, and the proteins are then detected by a color reaction. However, staining is quite often faint and difficult to interpret, and needs to be carefully compared to negative controls run in parallel.
Fig. 7.9 Immunoblot showing intrathecal synthesis of Hu antibodies. The band at 53 kDa, prominent at a serum dilution of 1:200 (IgG 50 mg/L), has the same intensity in serum and CSF when the IgG concentration is 5 mg/L in both, and thus represents a blood-derived antibody; whereas the triplet band between 35 kDa and 39 kDa, specific for anti-Hu, is now more pronounced in the CSF, thus indicating intrathecal synthesis.
Pitfalls
Recombinant proteins Depending on the degree of purification, recombinant proteins may retain a considerable fraction of proteins derived from E. coli or other organisms used for expression. These impurities may be responsible for weak falsepositive reactions.
Molecular weight data When evaluating the molecular weights of bands by comparison to those specified in the literature, it should be remembered that various differences may result from modifications of the SDS electrophoresis (reducing or nonreducing conditions, continuous gels or gradient gels) or simply from a skewed run. The kind of standard used and the method of evaluation (manual calculation or determination by special software) also affect the molecular weight data.
ELISA
An ELISA may be performed if recombinant onconeural antigens are available in highly purified form. The ELISA is mainly suited for quantitative follow-up (Rauer et al., 2002).
References
Benyahia B, Amoura Z, Rousseau A, et al. Paraneoplastic antineural antibodies in patients with systemic autoimmune diseases. J Neurooncol 2003;62:349–351
Bernal F, Shams'ili S, Rojas I, et al. Anti-Tr antibodies as markers of paraneoplastic degeneration and Hodgkin's disease. Neurology 2003;60:230–234
Budman CL, Kerjakovic M, Bruun RD. Viral infection and tic exacerbation [letter]. J Am Acad Child Adolesc Psychiatry 1997;36:162
Church AJ, Cardoso F, Dale RC, et al. Anti-basal ganglia antibodies in acute and persistent Sydenham's chorea. Neurology 2002;59:227–231
Dale RC. Autoimmunity and the basal ganglia: new insights into old diseases. Q J Med 2003;96:183–191
Darnell RB, Posner JB. Paraneoplastic syndromes involving the nervous system. N Engl J Med 2003;349:1543–1554
Greenlee JE, Boyden JW, Pingree M, et al. Antibody types and IgG subclasses in paraneoplastic neurological syndromes. J Neurol Sci 2001;184:131–137
Hadjivassiliou M, Davies-Jones GAB, Sanders DS, Grünewald RA. Dietary treatment of gluten ataxia. J Neurol Neurosurg Psychiatry 2003 a;74:1221–1224
Hadjivassiliou M, Williamson CA, Woodroofe NM. The humoral response in the pathogenesis of gluten ataxia. Neurology 2003 b;60:1397–1399
Honnorat J, Antoine J, Belin M. Are the “newly discovered” paraneoplastic anti-collapsin response-mediator protein 5 antibodies simply anti-CV2 antibodies? Ann Neurol 2001;50:688–690
Kaiser R. Paraneoplastische neurologische Syndrome. Diagnostische und pathogenetische Bedeutung von Autoantikörpern. Nervenarzt 1999;70:688–701
Lennon VA. Paraneoplastic autoantibodies: the case for a descriptive generic nomenclature. Neurology 1994;44:2236–2240
Llado A, Mannucci P, Carpentier AF, et al. Value of Hu antibody determinations in the follow-up of paraneoplastic neurological syndromes. Neurology 2004;63:1947–1949
Moll JW, Markusse HM, Pijnenburg JJ, et al. Antineural antibodies in patients with neurologic complications of primary Sjögren's syndrome. Neurology 1993;43:2574–2581
Murinson BB, Butler M, Marfurt K, et al. Markedly elevated GAD antibodies in SPS. Effect of age and illness duration. Neurology 2004;63:2146–2148
Pittock SJ, Lucchinetti CF, Lennon VA. ANNA-2: Paraneoplastic accompaniments. Ann Neurol 2003;53:580–587
Rauer S, Czygan M, Kaiser R. Quantification of circulating anti-Hu antibody in serial samples from patients with paraneoplastic neurological syndromes: possible correlation of antibody concentration and course of neurological symptoms. J Neurol 2002;249:285–289
Rees JH, Hain SF, Johnson MR, et al. The role of [18F]fluoro-2-deoxyglucose-PET scanning in the diagnosis of paraneoplastic neurological disorders. Brain 2001;124:2223–2231
Snider LA, Swedo S. Post-streptococcal autoimmune disorders of the nervous system. Curr Opin Neurol 2003;16:359–365
Voltz R. Paraneoplastische neurologische Erkrankungen. Nervenarzt 2002;73:909–929
Yu Z, Kryser AS, Griesmann GE, et al. CRMP-5 neuronal autoantibody: marker of lung cancer and thymoma-related autoimmunity. Ann Neurol 2001;49:146–154