Background Antibodies directed against glutamic acid decarboxylase (GAD) have been found in patients with a variety of neurological disorders including epilepsy. The presence autoantibodies in patients with epilepsy has provoked interest as it may suggest an autoimmune basis for the aetiology in some patients.

Methods In a tertiary epilepsy clinic we identified patients with adult onset epilepsy of unknown aetiology despite standard workup (including an epilepsy protocol MRI scan of the brain). These patients were tested for the presence of GAD antibodies (assay performed at the University of Oxford neuroimmunology laboratory). Patients were classified as being “negative”, “low positive” (antibody titre within the range defined by the laboratory) or “high positive” (antibody titre above this range).

The diagnostic yield was calculated and for the patients with high positive titres further clinical characteristics and therapeutic outcomes were recorded during the 2 year study period.

Results 112 patients were tested for GAD antibodies. 14 (12.5%) were positive with 6 (5.4%) of these being high positive.

Of the high positive patients, all were female with a mean age of 35 years. All had temporal lobe epilepsy. 1 patient had coexisting type 1 diabetes, 1 had rheumatoid arthritis and 1 had coeliac disease and pernicious anaemia. 1 patient developed optic neuritis during the study period and another had coexisting cerebellar ataxia. CSF analysis in all 6 showed oligoclonal bands and intrathecal GAD antibodies. MRI scans were normal in all cases, apart from one patient who had equivocal evidence of left hippocampal sclerosis (HS).

5 patients received immunotherapy (3 with steroids alone, 1 with intravenous immunoglobulin (IVIG) and azathioprine and 1 with steroids followed by plasma exchange, IVIG and azathioprine). No patient had a sustained improvement in seizure control. The patient with cerebellar ataxia experienced improvement with balance and coordination with steroid therapy. The patient with possible HS underwent temporal lobectomy, following further investigations including a PET scan, and has remained seizure free for 1 year. GAD antibodies have persisted in the serum postoperatively although histology confirmed HS with no evidence of inflammation.

Conclusion Our experience suggests that GAD antibodies are present in a significant number of patients with adult onset epilepsy, especially those with other autoimmune disorders.

The relevance of GAD antibodies in this setting remains unclear. The presence of OCBs and GAD antibodies within the CSF suggest a CNS based autoimmune process but the GAD antibodies themselves may be a marker of this processes rather than being directly pathogenic, as is suggested for antibodies directed against neuronal surface antigens. In our cohort immunotherapy did not improve seizure control and it may be that these antibodies are produced as part of a seizure related phenomenon.

Further work is required to evaluate the relevance of GAD antibodies detected in epilepsy patients.