Cerebellar Ataxia With Anti–Glutamic Acid Decarboxylase Antibodies: Title and subTitle BreakStudy of 14 Patients FREE

GLUTAMIC acid decarboxylase (GAD) is a major enzyme of the nervous system that catalyzes the conversion of glutamate to γ-aminobutyric acid (GABA). This enzyme, also expressed by pancreatic beta cells, has been identified as a dominant and essential autoantigen in the development of insulin-dependent (type 1) diabetes mellitus (IDDM).¹ Anti-GAD autoantibodies (GAD-Ab) are present in up to 80% of patients with newly diagnosed IDDM and can be detected many years before the clinical onset of the disease.² High levels of GAD-Ab are found in the serum and cerebrospinal fluid (CSF) of at least 60% of patients with the stiff-man syndrome, a rare disorder of the central nervous system characterized by progressive muscle rigidity with superimposed painful spasms.^{3 - 4} Patients with stiff-man syndrome who have GAD-Ab usually also have IDDM and other organ-specific autoimmune manifestations, suggesting that the stiff-man syndrome may have an autoimmune-mediated pathogenesis.⁵ The GAD-Ab of patients with stiff-man syndrome are in higher titer and have a different epitope specificity than those of patients with IDDM, suggesting that the 2 diseases may have different autoimmune mechanisms.^{6 - 7}

Recent studies show that high levels of GAD-Ab may also be observed in a few patients with cerebellar ataxia, supporting an autoimmune pathogenesis of the cerebellar syndrome.^{8 - 14} To confirm this clinical-immunologic association, we describe a series of 14 patients with idiopathic cerebellar ataxia and high titers of GAD-Ab.

Serum samples were selected from those sent to 4 European laboratories (Barcelona, Spain; Lyon and Saint-Etienne, France; and Padua, Italy) to test for the presence of antineuronal autoantibodies associated with paraneoplastic neurologic syndromes. The selection criterion was an immunohistochemical pattern compatible with GAD-Ab. The presence of GAD-Ab was confirmed by radioimmunoassay (RIA) in 2 centers. Thirty-six serum samples from more than 9000 showed an immunohistochemical pattern compatible with GAD-Ab; 22 of them were from patients with typical stiff-man syndrome and the other 14 had ataxia as the predominant neurologic symptom. None of the selected serum samples had anti-Hu, anti-Yo, anti-CV2, anti-Tr, or anti-Ri antibodies The clinical features of these 14 patients were obtained by personal clinical interview and neurologic examination, to establish the presence of a personal or familial history of autoimmune disorders and evolution of ataxia. Samples of CSF were obtained from all patients for standard examination and the presence of IgG oligoclonal bands. In 6 patients, paired serum-CSF samples were available to analyze a possible intrathecal synthesis of GAD-Ab.

To standardize the immunologic data, serum and, when available, CSF samples of the 14 patients were reevaluated by the techniques described below. The samples detected by immunohistochemical analysis were sent to the different laboratories with positive and negative controls (20 with idiopathic late-onset cerebellar ataxia and no anti-GAD reactivity by immunohistochemical analysis; 9 with polyendocrine autoimmunity [IDDM and thyroiditis] without neurologic symptoms and selected for presenting high GAD-Ab levels by RIA; 91 healthy subjects; 10 patients with stiff-man syndrome; and 49 with recent diagnosis of IDDM) and tested without awareness of the clinical diagnosis.

Immunohistochemical analysis was performed on frozen sections of paraformaldehyde-fixed rat cerebellum by means of an avidin-biotin immunoperoxidase (Barcelona's laboratory) and indirect immunofluorescence technique (Lyon's laboratory) as previously described.^{10 ,12} Intrathecal synthesis of GAD-Ab was determined by means of the reciprocals of end-point titers determined by immunohistochemical analysis of paired serum-CSF samples and applying the Schüller formula as previously described.¹² A ratio greater than 2 was considered positive for intrathecal synthesis.

Immunoblots were done in Barcelona with a technique previously described.¹² Briefly, human GAD65 recombinant protein (CIS Biointernational, Gif-sur-Yvette, France) (0.16 µg per lane) was electrophoretically separated in a 10% sodium dodecyl sulfate–polyacrylamide gel and transferred to nitrocellulose.¹⁵ Strips were sequentially incubated with the patient's serum or GAD-6 monoclonal antibody (Hybridoma Bank, Iowa City, Iowa), immunoreacted with an avidin-biotin technique, and developed with diaminobenzidine tetrahydrochloride (Sigma-Aldrich Corp, St Louis, Mo).

To confirm the anti-GAD specificity of the serum samples that were positive on immunohistochemical analysis, RIAs were performed with GAD labeled with iodine 125 (¹²⁵I), obtained from 1 of 2 sources (CIS Biointernational or RSR Ltd [Cardiff, Wales]). In the Barcelona laboratory ¹²⁵I-GAD was used according to the manufacturer's instructions^{12 ,16} ; in the Oxford, England, laboratory slight modifications were made as previously described.¹⁷

Serum samples from the 14 patients with ataxia who had GAD-Ab and the 20 patients with idiopathic late-onset cerebellar ataxia and no anti-GAD reactivity by immunohistochemical analysis were tested in the Lyon laboratory for the presence of other organ-specific autoantibodies. Antinuclear, antiendomysium, antiadrenal, antipituitary, and antistomach autoantibodies were detected by an indirect immunofluorescence technique as previously described.¹⁸ Antithyroperoxidase, antithyroglobulin, antigliadin, and anti–parietal cell autoantibodies were detected by enzyme-linked immunosorbent assay commercial kits according to the instructions of the manufacturers (Pharmacia Upjohn Diagnostics, Uppsala, Sweden, and Sigma Chemical Co, Paris, France). The IA₂ antibodies were detected by radioligand assay.¹⁹

The main clinical features of the 14 patients with GAD-Ab and cerebellar ataxia are summarized in Table 1 and Table 2 and include 4 patients (patients 1-4) who were previously described.^{10 - 12} Thirteen of them were women, with a median age of 51 years (range, 20-74 years) at the onset of the cerebellar syndrome. Six patients had a family history of autoimmune diseases such as IDDM or thyroiditis, but none had a family history of cerebellar ataxia. Ten of 14 patients had IDDM. In all cases, these patients had a late-onset IDDM that started at a median age of 47 years (range, 28-66 years). The diagnosis of IDDM antedated the onset of the cerebellar syndrome in 7 patients (median, 8 years; range, 2-12 years). Other autoimmune disorders were thyroiditis (Hashimoto or Graves disease) (8 patients), pernicious anemia (2 patients), myasthenia gravis (1 patient), and psoriasis (1 patient). An asymptomatic celiac disease was diagnosed by duodenal biopsy in 1 patient after antigliadin autoantibodies were detected by a systematic test. Two patients presented with thymoma 3 years before and 2 years after the diagnosis of the cerebellar ataxia.

The cerebellar ataxia that developed in these 14 patients was similar (Table 2). A subacute onset was observed in only 1 case, whereas symptoms progressed slowly, suggesting a degenerative disease, in 13 patients. The main cerebellar sign, which was present in all of the patients, was gait ataxia that was moderate or severe in 10. Limb ataxia was also observed in 12 patients, but was rated as mild in 7. The cerebellar symptoms prevented a completely independent way of life (Rankin score, >2) in 10 patients. Nystagmus was observed in 12 patients and dysarthria in 8. Associated neurologic symptoms were leg rigidity suggesting a focal stiff-man syndrome (2 patients), peripheral neuropathy (1 patient), and myasthenia gravis (1 patient). The CSF examination gave normal results in only 4 patients. For the 10 other patients, CSF analysis showed a normal protein level, but isoelectric focusing and immunoblot of IgG detected oligoclonal IgG bands not present in the serum. Three patients had a high IgG index. Only 1 patient had an abnormal cell count (7 lymphocytes). Magnetic resonance images of the brain were normal in 7 patients and showed pure cerebellar atrophy in the others. One patient had an associated Arnold-Chiari type I malformation (patient 4), but none showed brainstem atrophy.

The RIA confirmed anti-GAD reactivity observed by immunohistochemical analysis in all 14 serum samples by the 2 laboratories with complete agreement. A positive intrathecal synthesis of GAD-Ab was observed in 5 of the 6 patients who had paired serum-CSF samples available (median, 11.5; range, 2.9-16.4).

All 14 serum samples of patients with cerebellar ataxia had a high level of GAD-Ab by RIA (Figure 1). The mean (± SD) level of GAD-Ab in these patients (37 300 ± 30 460 U/mL) was similar to the level in those with stiff-man syndrome (21 750 ± 19 280 U/mL) and significantly higher than the level in those with a recent diagnosis of IDDM (48 ± 112 U/mL) or with polyendocrine autoimmunity without neurologic disease (5828 ± 9355 U/mL) (P<.001). However, in 6 of the 9 patients with polyendocrine autoimmunity, serum GAD-Ab levels overlapped with those of patients with cerebellar ataxia and GAD-Ab (Figure 1). None of the control patients with idiopathic late-onset cerebellar ataxia without GAD reactivity by immunohistochemical analysis or the normal subjects had GAD-Ab by RIA.

The 14 serum samples displayed a band in immunoblots of GAD65 recombinant protein with the same electrophoretic mobility as that recognized by GAD-6 monoclonal antibody (Figure 2). A similar band was also observed with 7 of the 9 serum samples from patients with polyendocrine autoimmunity without neurologic disease (data not shown). Other organ-specific autoantibodies were detected in 11 of the 14 patients with cerebellar ataxia and GAD-Ab (Table 1) and only in 2 of the 20 with idiopathic late-onset cerebellar ataxia without GAD-Ab (P<.01). The 2 patients showed antithyroperoxidase antibodies, but none of them had antigliadin or antiendomysium antibodies.

This study describes a series of patients with cerebellar ataxia and high titers of GAD-Ab. It is important to recognize this subgroup of patients because they may respond to immunotherapies. Previous case reports of cerebellar ataxia and high titers of GAD-Ab did not establish a clear relationship between the presence of GAD-Ab and the cerebellar ataxia.^{8 - 14} The presence in most of these patients of oligoclonal bands in the CSF, GAD-Ab intrathecal synthesis, and late-onset IDDM and other organ-specific autoimmune disorders clearly distinguishes them from those with other nonfamilial late-onset cerebellar ataxia.

All but 1 patient presented with an insidious cerebellar ataxia without brainstem involvement that would fit in the first group of Harding's classification on idiopathic late-onset cerebellar ataxia²⁰ and that corresponds to the old Marie-Foix-Alajouanine type of cerebellar degeneration.²¹ We did not observe clinical or radiologic evidence of brainstem involvement, suggesting in our patients an olivopontocerebellar atrophy. Two patients had rigidity of one leg not explained by corticospinal tract dysfunction. These patients probably had a focal form of stiff-man syndrome.²² The co-occurrence of stiff-man syndrome and cerebellar ataxia with high titer of GAD-Ab was also described in 2 other patients in the literature.^{11 ,23} These facts suggest an overlap between the 2 neurologic symptoms and further emphasize the association of GAD-Ab with both cerebellar ataxia and stiff-man syndrome and suggest common pathologic mechanisms.

The clinical and serologic data of our 14 patients favor an immune system dysregulation. All but 1 patient were women, and many had other organ-specific, mostly endocrine, autoimmune manifestations. Indeed, all but 2 patients had evidence of other autoimmune disorders. Late-onset IDDM was observed in 71% of patients, thyroiditis in 57%, and a family history of autoimmune disorders in 43%. Furthermore, 2 patients developed a thymoma that is commonly associated with autoimmune neurologic or hematologic disorders. This clinical and immunologic profile is similar to that observed in stiff-man syndrome with GAD-Ab, in which 30% have IDDM.⁵ Finally, other organ-specific autoantibodies were observed in 79% of our patients, compared with only 10% of our control patients with cerebellar ataxia but without GAD-Ab.

Two of our patients with GAD-Ab also had serum IgA antigliadin antibodies. These antibodies have recently been described in patients with late-onset cerebellar ataxia, gluten sensitivity, and appropriate HLA genotype for celiac disease.²⁴ However, patients with gluten ataxia differ from those with GAD-Ab by the preponderance of males, high frequency of peripheral neuropathy, and low prevalence of nystagmus.^{24 - 25} Therefore, the presence of antigliadin antibodies in 2 of our patients probably reflects the coincidence with other autoimmune disorder without direct relation to the cerebellar ataxia.

Although there is evidence of immune dysregulation in our group of patients, the pathogenic role of autoimmunity and GAD-Ab in the cerebellar ataxia remains unclear. The GAD-Ab could merely reflect the presence of IDDM and polyendocrine autoimmunity observed in our patients. In fact, our selected control patients with polyendocrine autoimmunity without neurologic disease demonstrate that a high titer of GAD-Ab in the serum does not necessarily indicate an association with a given neurologic disorder. However, this possibility would not apply to the 4 patients who did not have IDDM, and our patients showed elevated intrathecal synthesis of GAD-Ab and positive oligoclonal IgG bands in the CSF. Both features indicate an active immune process in the nervous system that should not be expected if the cerebellar ataxia had a degenerative cause.

Recent work suggested that the GAD-Ab from patients with stiff-man syndrome could be pathogenic because, unlike the GAD-Ab from patients with IDDM, they can reduce GAD enzyme activity and GABA synthesis.²⁶ In 1 case of cerebellar ataxia, the GAD-Ab were able to cause a selective suppression of GABA-ergic transmission in in vitro experiments with isolated rat cerebellar slices.¹⁴ These studies, however, do not explain how GAD-Ab can access the intracytoplasmic antigen and why their actions are so selective compared with the widespread distribution of GAD in the nervous system. An alternative explanation could be that the high anti-GAD titers reflect only the presence of a more complex immune reaction against the nervous system that could implicate a cell-mediated immune response directed against GAD or other cerebellar antigens. Even if GAD is not the key antigen, the immunologic profile of these patients and their predisposition to develop organ-specific autoimmune manifestations, coupled with the presence of oligoclonal IgG bands in the CSF, suggests an immune-mediated disorder as the most probable pathogenesis of the cerebellar dysfunction.

In conclusion, our findings support a link between GAD autoimmunity and some cases of idiopathic cerebellar ataxia. We suggest that testing for GAD-Ab may be indicated in patients with cerebellar ataxia, particularly women with late-onset IDDM or other organ-specific autoimmune diseases and absence of brainstem involvement. If high levels of GAD-Ab are detected in the serum, the CSF should be examined for oligoclonal IgG bands and intrathecal synthesis of GAD-Ab. If CSF studies are positive, immune-mediated damage of the cerebellum linked to GAD autoimmunity is highly probable and immunotherapy could be warranted.

Accepted for publication October 2, 2000.

This study was supported in part by grant SGR 9500027, Generalitat de Catalunya, Barcelona, Spain; Institut National de la Santé et de la Récherche Médicale, Paris, France; La Fondation de France, Paris; La Ligue contre le Cancer du Rhône, Lyon, France; l'Association pour la Recherche contre le Cancer, Paris; and grant 756/01/97, Regione Veneto, Venice, Italy.

We thank Mercé Bonastre and Michelle Aguera for their excellent technical assistance and Isabelle Serre, MD, Caroline Tilikete, MD, and Bernard Carlander, MD, who provided some clinical data.

Corresponding author and reprints: Francesc Graus, MD, Service of Neurology, Hospital Clínic, Villarroel 170, 08036 Barcelona, Spain (e-mail: graus@medicina.ub.es).