GAD antibodies in neurological disease: a critical evaluation of the utility and treatment implications of GAD antibodies in clinical practice.

BACKGROUND

The interpretation of antibodies to glutamic acid decarboxylase 65 (GAD-Abs) in neurological practice is challenging. GAD-Abs are not considered directly pathogenic and immunotherapy guidelines are lacking.

METHODS

We undertook a single-center retrospective service evaluation of GAD-Abs, documenting clinical features, immunotherapy responses, and outcomes of 335 patients with positive GAD-Abs measured by indirect ELISA between 2012 and 2020. The serum:CSF ratio of GAD-Ab values was used as a surrogate for intrathecal synthesis.

RESULTS

168 (50%) patients had diagnosed neurological disorders (GAD-ND). Ninety-six had neurological disorders often or sometimes associated with GAD-Abs, i.e., stiff person syndrome spectrum disorders (SPS-SD, n = 26), cerebellar ataxia (n = 21), epilepsy (n = 19), encephalitis (n = 18), or any combination of these ("mixed", n = 12). Seventy-two had other neurological disorders (ONDs) not typically associated with GAD-Abs. We defined a cut-off of 10,000 IU/mL a priori and a posteriori for GAD-Ab associated NDs, but identified values > 10,000 IU/mL in 21% and 11% of patients with ONDs or diabetes respectively, and < 10,000 IU/mL in 39% patients with classical GAD-Ab syndromes, indicating low assay specificity and sensitivity. Low serum: CSF GAD-Ab ratios were consistent with intrathecal synthesis in 12/19 tested; 25/54 patients had oligoclonal bands. 30/50 patients given adequate immunotherapies had partial (n = 17) or good (n = 13) responses, particularly those with SPS-SD or limbic encephalitis. Within the limitations of small subgroups and routine laboratory titrations, patients with GAD-Ab values > 10,000 IU/mL, intrathecal synthesis of GAD-Abs, or oligoclonal bands, were not more likely to improve with immunotherapies than those with GAD-Ab values < 10,000 IU/mL and a non-inflammatory CSF. Rather, treatment response correlated with disease group, principally SPS-SD and encephalitis.

CONCLUSIONS

These results suggest caution in over-interpreting GAD-Abs values. Better biomarkers for identifying patients with immunotherapy responsive GAD-Ab disease are needed.