Autoantibodies to glutamic acid decarboxylase in three patients with cerebellar ataxia, late-onset insulin-dependent diabetes mellitus, and polyendocrine autoimmunity

Glutamic acid decarboxylase (GAD), the enzyme that catalyzes the conversion of glutamate to GABA, has been identified as a dominant autoantigen in two diseases, stiff-man syndrome (SMS)^1,2 and insulin-dependent diabetes mellitus (IDDM).^3,4 GAD autoantibodies (GAD-Abs) are present in approximately 80% of newly diagnosed IDDM patients and can be detected many years before the clinical onset of the disease.³ SMS is a rare disorder of the CNS characterized by progressive muscle rigidity with superimposed painful spasms.⁵ Around 60% of patients with SMS harbor GAD-Abs in their serum and CSF. SMS patients with GAD-Abs usually present IDDM and other organ-specific autoimmune manifestations, suggesting that SMS may have an autoimmune-mediated pathogenesis.⁴ GAD-Abs from patients with SMS have higher titers and a different epitope specificity compared with those of patients with IDDM.^3,6,7

GAD-Abs are reported in a few patients with degenerative diseases of the cerebellum,^2,8-12 palatal myoclonus,¹³ and parkinsonism.^10,12 However, all but two patients^11,13 were included in a varied control group of degenerative diseases used to ascertain the specificity of GAD-Abs for SMS, and a clinical-immunologic correlation was not done.

We report GAD-Abs in three patients with cerebellar ataxia and late-onset IDDM, in the setting of organ-specific autoimmunity similar to that observed in most SMS patients with positive GAD-Abs.

Case reports. Patient 1. A 76-year-old woman was admitted in December 1994 with a 15-month history of progressive gait disturbance and dysarthria. She was diagnosed with diabetes mellitus 10 years earlier that has required insulin for the past 5 years. The family history was unremarkable, except that her mother had IDDM. The general examination disclosed cheilitis and depapillated tongue. On neurologic examination, there was marked dysarthria, impairment of smooth-pursuit eye movements, limb dysmetria, intention tremor, and severe ataxia of stance and gait. The rest of the neurologic examination was normal.

Magnetic resonance imaging showed diffuse cerebellar atrophy without involvement of the brainstem. Needle EMG and nerve conduction velocity studies were normal. Pernicious anemia was diagnosed based on a low serum vitamin B₁₂ of 137 pg/mL (normal, 200 to 1,000 pg/mL) with a positive Schilling test and gastric atrophy demonstrated by gastroscopy and biopsy. There was an elevated level of thyroid-stimulating hormone (34.26µIU/mL) with normal level of serum thyroid hormones, suggestive of sub-clinical hypothyroidism. Echography of the thyroid showed a multinodular goiter. The following organ-specific autoantibodies were detected: pancreatic islet-cell, intrinsic factor, thyroglobulin, and thyroid microsomal antibodies. Adrenal antibodies were negative. The CSF analysis was normal except for an IgG index of 1.5 (upper normal value, 0.7) and oligoclonal IgG bands, not present in the serum, evaluated by isoelectric focusing and immunoblot of IgG.

The patient was treated with cyanocobalamin and L-thyroxin without improvement of the cerebellar syndrome. In the last 18 months she presented two generalized seizures that required treatment with phenytoin.

Patient 2. A 60-year-old woman noticed gait instability over 2 weeks in December 1995 that gradually worsened in the ensuing 3 months. Graves' disease was diagnosed at age 48 years, and diabetes mellitus, which required insulin treatment 1 month after diagnosis, was diagnosed 5 years later. Her family history was remarkable for late-onset IDDM in two sisters; one also has vitiligo. General examination was normal except for goiter. The neurologic examination disclosed dysarthria, bilateral horizontal nystagmus, limb dysmetria, intention tremor, and severe truncal ataxia. Neurologic examination was otherwise normal.

MRI showed mild atrophy of the cerebellum. EMG and the level of thyroid hormones were normal. The following organ-specific autoantibodies were detected: pancreatic islet-cell, thyroid microsomal, and gastric parietal cell antibodies. Thyroglobulin and adrenal antibodies were negative. The CSF examination was normal except for the presence of oligoclonal IgG bands, absent in the serum, evaluated by isoelectric focusing and immunoblot of IgG. HLA typing was as follows: DRB1*03,07,B3,B4,DQB1*02. The cerebellar disorder remained unchanged in September 1996.

Patient 3. A 52-year-old woman was evaluated in February 1990 because of progressive unsteadiness of gait, vertigo, and oscillopsia related to postural head changes during the last year. Her medical history was significant for diabetes mellitus diagnosed at age 45 years that required insulin treatment at 6 months after diagnosis as well as psoriasis at age 38 years. There was a family history of late-onset IDDM (father and two paternal uncles). General examination was normal except for psoriasis. The neurologic examination disclosed only downbeat vertical nystagmus and ataxic gait. MRI showed type I Arnold-Chiari malformation. Over the ensuing 2 years, she noticed progressive dysarthria and her gait ataxia. Clinical progression was believed to be related to the Arnold-Chiari malformation, and the patient underwent posterior suboccipital decompression in June 1992. After surgery, her dysarthria and downbeat nystagmus improved but not her gait. Over the ensuing 3 years she developed clumsiness of the right hand and increasing unsteadiness. Neurologic examination in September 1996 revealed mild dysarthria, upward vertical nystagmus, limb dysmetria more marked on the right, and severe ataxic gait that prevented her from walking alone more than a few meters. MRI demonstrated resolution of cerebellar tonsillar descent and mild atrophy of the vermis. Routine laboratory analysis, including thyroid hormones, were normal. The following organ-specific autoantibodies were detected: pancreatic islet-cell, gastric parietal cell, and thyroid microsomal antibodies, whereas adrenal and thyroglobulin antibodies were negative. The CSF analysis was normal except for oligoclonal IgG bands, not present in the serum, evaluated by isoelectric focusing and immunoblot of IgG. HLA typing was as follows: DRB1*01,12,B3,DQB1*0301,0501.

Methods. Sera. Serum and CSF samples were obtained from the three patients. Patients 3 had serum samples from 1990 and 1996. We compared the levels of GAD-Abs of these three patients with those of (1) five patients (median age, 61 years; range, 54 to 75 years) with the diagnosis of SMS, according to accepted criteria⁵; (2) 49 patients with IDDM, 34 with newly diagnosed IDDM (median age, 29 years; range, 11 to 78 years) and nine with thyroid antibodies; and (3) 64 patients with late-onset cerebellar ataxia of probable degenerative origin, 10 of whom had familial olivo-ponto-cerebellar atrophy (median age, 61 years; range, 29 to 81 years)(median duration of disease, 3 years; range, 1 to 20 years) (five of the 64 patients had type II diabetes mellitus but none had IDDM or clinical features of polyendocrine autoimmunity or other autoimmune disorder); (4) 14 non-IDDM islet-cell antibody-positive first-degree relatives of IDDM patients (median age, 18 years; range, 10 to 50 years); and (5) 91 healthy blood donors(median age, 26 years; range, 19 to 73 years) with no family history of IDDM.

Anti-glutamic acid decarboxylase autoantibodies. The presence of GAD-Abs was assessed in all serum samples by radioimmunoassay and immunohistochemistry, and the positive samples were confirmed by immunoblot.

Radioimmunoassay. GAD-Abs were measured with a commercial kit(CIS Biointernational, Gif-sur-Yvette France) following the manufacturer's instructions.¹⁴ Briefly, 20 µL of standards (SMS serum containing GAD-Abs at different dilutions expressed in arbitrary units, U/mL) and serum samples were incubated with 50 µL of¹²⁵ I-labelled human recombinant GAD65 for 2 hours at room temperature. Then, 50 µL of protein A-sepharose was added and the mixture incubated for 1 hour at room temperature. After centrifugation at 1,500g for 30 minutes at 4°C, the precipitates were counted for ¹²⁵I with a gamma scintillation counter. The results were interpolated in the standard curve constructed using the dilutions of the positive control serum. For each serum, conditions were established in that appropriate dilutions of the serum produced results that were in the straight line portion of the standard curve. All samples were tested in duplicate. The inter- and intra-assay coefficient of variation was below 10%.

Immunohistochemistry. Wistar rats were anesthetized and perfused with saline followed with 4% paraformaldehyde in phosphate buffered saline (PBS). The cerebellum was further fixated with 4% paraformaldehyde for 4 hours and cryoprotected with 20% sucrose in PBS overnight. Ten µm frozen sections were air dried and, after inhibition of endogenous peroxidase with 0.3% hydrogen peroxide in PBS for 10 minutes, were sequentially incubated with 10% normal goat serum for 20 minutes, patient's serum(screening dilution 1:500) for 3 hours at 37°C, biotinylated goat anti-human IgG for 30 minutes, and the avidin biotin immunoperoxidase complex(Vector Labs, Burlingame, CA) for 30 minutes. The reaction was developed with 0.05% diaminobenzidine tetrahydrochloride (Sigma Chemical Co., St. Louis, MO) with 0.01% hydrogen peroxide in PBS with 0.5% Triton X-100. Dilution of antibodies was done in PBS with 0.3% Triton X-100.

Immunoblot. Human GAD65 recombinant protein (CIS Biointernational) (0.16 µg/lane) was electrophoretically separated in a 10% sodium dodecyl sulfate-polyacrylamide gel and transferred to nitrocellulose.¹⁵ After blocking with 5% dry Carnation milk and 10% normal goat serum, strips were incubated with the patient's serum (1:1,000 dilution) or GAD-6 monoclonal antibody (Hybridoma Bank, Iowa City, IA) for 12 hours at room temperature, washed with PBS, and incubated with biotinylated goat anti-human IgG, diluted 1:2,000, or horse antimouse IgG, diluted 1:1,000 in 10% normal goat serum for 1 hour. Strips were then processed as described in the immunohistochemistry technique.

Intrathecal synthesis of GAD-Abs. Intrathecal (IT) GAD-Ab synthesis was calculated by Schüller's formula as previously described.¹⁶ Briefly, the GAD-Ab-specific activity(ASA) was estimated in the serum and CSF by the following formula: ASA = serum or CSF titer/[IgG] (mg/L) and expressed in units 10^-3 (using the reciprocals of end point titers determined by immunohistochemistry). The IT GAD-Ab synthesis was calculated by the following formula: IT ASA = (CSF ASA-[serum ASA × percentage of IgG from serum])/percentage of IT IgG.

A ratio of IT ASA/serum ASA of GAD-Abs >2 was considered a positive IT synthesis.¹⁶

Results. The serum and CSF of the three patients recognized a band in immunoblots of GAD65 recombinant protein with the same electrophoretic mobility as that recognized by GAD-6 monoclonal antibody or serum of SMS patients (figure 1). By radioimmunoassay, the serum of the three patients had levels of GAD-Abs of 9,300 U/mL, 39,500 U/mL, and 14,000 U/mL that were similar to those of patients with SMS (median, 20,500 U/mL; range, 600 to 67,000 U/mL). The GAD-Ab levels were significantly lower in IDDM patients (48 ± 112 U/mL) (p < 0.0001) and ICA-positive relatives (170 ± 203 U/mL) (p < 0.0001). None of the 64 patients with cerebellar ataxia not associated with autoimmune disease and the normal subjects had GAD-Abs (figure 2). In the immunohistochemistry studies, only the serum and CSF samples from the three patients, and all but one serum sample of the five SMS patients, showed immunostaining of GABA-ergic nerve terminals on rat cerebellar tissue sections(figure 3). Serum GAD-Ab titers by immunohistochemistry were 1:20,000, 1:16,000, and 1:8,000 in the three patients, and the corresponding CSF titers were 1:400, 1:160, and 1:400. The ratio of IT ASA/serum ASA of GAD-Abs was 16.4, 2.9, and 13.5, respectively, consistent with a positive IT synthesis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Immunoblots of GAD 65 recombinant protein probed with the serum of patients with cerebellar ataxia and autoimmune features (lanes 1 to 3), SMS (lanes 4 and 5), and a normal subject(lane 6). The immunoreactive bands have the same electrophoretic mobility as that recognized by GAD-6 monoclonal antibody (lane G).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Levels of GAD-Abs, by radioimmunoassay, expressed in arbitrary units, U/mL. Note that the y-axis is a log scale. Along the x-axis: 1 = Control (n = 91); 2 = Cerebellar Ataxia (n = 64); 3 = IDDM (n = 49); 4 = Relatives of IDDM (n = 14); 5 = "Autoimmune" Cerebellar Ataxia (n = 3); and 6 = SMS (n = 5).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Section of rat cerebellum incubated with the serum from patient 1 showing the characteristic immunoreactivity of GAD-Abs. Counterstained with hematoxylin, magnification ×400 before 22% reduction.

Discussion. The observation of three patients with cerebellar ataxia, high-titer GAD-Abs, and similar clinical and serologic features suggests that GAD-Abs may be related to cerebellar dysfunction and extend the spectrum of neurologic disorders associated with GAD autoimmunity.

GAD-Abs have been detected in a few patients with neurologic disorders other than SMS.^2,8-13 Honorat et al.¹¹ described a patient with cerebellar cortical atrophy, peripheral neuropathy, and slow saccadic eye movements and GAD-Abs. The patient was not diabetic and did not have other features of polyendocrine autoimmunity. The other patients with cerebellar ataxia and GAD-Abs (two patients with late-onset cerebellar ataxia and another with multisystem atrophy) were described in abstract form^8,9 and were included as part of a "control" group of neurologic disorders in a study that evaluated the frequency of GAD-Abs in patients with SMS.^2,10 These patients were diabetic, but no additional information was provided. In addition, two recent reports have emphasized the presence of a cerebellar disorder at the same time or after the diagnosis of SMS.^17,18

Our patients presented with an isolated cerebellar syndrome. In one of them, the onset was subacute, with a clinical pattern similar to that of paraneoplastic cerebellar degeneration.¹⁹ The other two patients did not develop clinical or radiologic involvement of the brainstem during the follow-up for more than 3 years, and the course of the cerebellar syndrome was not different from that observed in idiopathic late-onset cerebellar ataxia.²⁰ The three patients had several organ-specific, mostly endocrine, autoimmune manifestations, and presented an HLA genotype associated with the susceptibility to develop autoimmune disorders.^21,22 All of them had a family history of IDDM, in contrast to the 20% frequency of IDDM found in first-degree relatives of patients with this disease.²³ This clinical and immunologic profile is common to that observed in SMS with GAD-Abs. IDDM, usually of late onset, occurs in up to 30% of patients with SMS.⁴ Epilepsy, present in patient 1, is also reported in about 10% of SMS patients.²

It is unlikely that the GAD-Abs found in our patients were only related to their IDDM because GAD-Ab levels were higher and did not overlap with those of IDDM patients without neurologic disorders. Moreover, the presence of a positive IT synthesis of GAD-Abs in the CSF samples would not be expected if GAD-Abs were just related to the IDDM.

Could GAD-Abs be responsible for the cerebellar dysfunction of our patients? GAD-Abs were not detected in patients with cerebellar ataxia without the autoimmune background of the GAD-Ab-positive patients. In addition, with the immunohistochemistry technique described in this study that can detect high GAD-Ab titers, we did not find another positive serum sample among more than 300 taken from patients with several neurologic disorders studied in our laboratory (Graus, unpublished observation). These data further suggest high GAD-Ab titers are unlikely to be produced as a consequence of the damage of GABA-ergic neurons by other causes.

Although GAD is a cytoplasmic antigen, there is experimental evidence that shows Purkinje cells may uptake IgG from the CSF.²⁴ Then, GAD-Abs that are present in high levels in the CSF of these patients could theoretically be internalized by the Purkinje cells in vivo, interact with GAD, and cause the neuronal dysfunction. An alternative explanation is that the high GAD-Ab titers only reflect the presence of a more complex, probably cell-mediated immune response against GAD or the predisposition of these patients to a multiantigenic autoimmunity. Even if GAD is not the key antigen, the immunologic profile of these patients and the presence of oligoclonal IgG bands in the CSF suggest that the most probable pathogenesis of the cerebellar disorder is an immune response to a still unknown Purkinje cell antigen, and a trial of immunosuppressor therapy may be warranted.

Other possible causes for the cerebellar syndrome were reasonably excluded in our patients. None presented alcohol abuse, serum anti-Yo antibodies, or an underlying tumor. Cerebellar ataxia has been described in patients with myxedema or pernicious anemia.²⁵ Nevertheless, patient 1 did not improve with vitamin B₁₂ or thyroid replacement, and patients 2 and 3 were euthyroid.

In conclusion, our findings suggest a possible link of GAD autoimmunity not only with SMS, but also with at least some cases of cerebellar ataxia. We suggest that testing for GAD-Abs may be indicated in patients with cerebellar ataxia associated with late-onset IDDM especially if they have a family history of IDDM or present clinical or serologic evidence of other organ-specific autoimmune manifestations.

Acknowledgments

We are indebted to Mercè Bonastre, Eva Sanchez, and Rosa Pagés for their technical assistance and to Drs. R. Gomis and C. Rodriguez for their valuable comments on the manuscript. The GAD 6 monoclonal antibody was obtained from the Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA.