A man aged 19 years, previously in good health with no history of febrile convulsions or head injury or family history of diabetes or other autoimmune disorders, suddenly developed complex partial seizures. The episodes occurred several times a day over the next 4 months. Routine blood analyses and neurological assessments were normal. Electroencephalography showed temporal and parietal spikes, and signal abnormalities were seen on magnetic-resonance images. Analysis of cerebrospinal fluid showed mild lymphocytic pleiocytosis (25/mm3), increased IgG intrathecal synthesis (IgG index 0· 8; normal< 0· 7) and IgG oligoclonal bands, which confirmed acute encephalitis. PCR analysis for herpesvirus and antibodies, and antibodies to Epstein-Barr virus and to cytomegalovirus were negative in the serum and cerebrospinal fluid. Thoracic and abdominal computedtomography scans did not show tumours at 1 year follow-up. Multiple antiepileptic drugs were tried with no success, but transient remission (2 weeks) was obtained with methylprednisolone (1 g daily for 5 days). On 4% paraformaldehyde fixed rat cerebellum sections, the patient’s serum and cerebrospinal fluid stained typically glutamic acid decarboxylase (GAD)-rich producing neurons (titre 1: 64000 in serum; 1: 1500 in the cerebrospinal fluid). Three serum samples of patients with Stiffman syndrome with GAD autoantibodies gave the same pattern. On western blot with human neuronal proteins, the serum recognised a 65 kDa-band equivalent to the molecular weight of GAD and was positive for human recombinant GAD65 by radioimmunoprecipitation (320 LH-u and 1100 LH-u in two samples taken 5 months apart). The serum samples also gave a selective staining on pancreatic islets, similar to that for sera which are positive only for GAD autoantibodies. 1 The antibody-specificity index was 5· 07, indicative of intrathecal synthesis of GAD autoantibodies. With the exception of antinuclear (1: 1600 and 1: 1280 diffuse pattern in two serum samples, respectively) and antidouble-stranded DNA (1: 40) antibodies, no other islet-related, organ-specific and nonorgan-specific autoantibodies were detected. About 60% of patients with Stiffman syndrome possess GAD autoantibodies, but are at higher titre and with different GAD eiptope specificity, compared with the GAD autoantibodies of patients with type 1 diabetes. 2 A proportion of patients with Stiffman syndrome with GAD autoantibodies have type 1 diabetes and other organ-specific autoimmunity. Such patients are usually females, but our patient was a male with no other islet-related or organ-specific antibodies (only antinuclear antibodies were found). GAD autoantibodies have also been described in patients with cerebellar ataxia3 and in a case with palatal myoclonus and epilepsy. 4 Whether GAD autoantibodies contribute to the pathogenesis of these disorders remains to be fully substantiated, but the beneficial effects of plasmapheresis in Stiffman syndrome is notable. 5 We have shown that an autoimmune temporal-lobe epilepsy could be associated with GAD autoantibodies. We recommend testing for the presence of GAD autoantibodies in patients with drug-refractory temporal-lobe epilepsy (15–20% of all epilepsy cases) to identify a potentially treatable subgroup, as in Stiffman syndrome. Further studies, including the characterisation of immunodominant epitope (s) on the GAD molecule, are needed to identify the neurological manifestations in patients with GAD autoantibodies.

1 Genovese S, Bonifacio E, McNally JM, et al. Distinct cytplasmic islet cell antibodies with different risks for Type I (insulin dependent) diabetes mellitus …