IMPORTANCE

Autoimmune epilepsy is an underrecognized condition, and its true incidence is unknown. Identifying patients with an underlying autoimmune origin is critical because these patients' condition may remain refractory to conventional antiseizure medications but may respond to immunotherapy.

OBJECTIVE

To determine the prevalence of neurological autoantibodies (Abs) among adult patients with epilepsy of unknown etiology.

DESIGN, SETTING, AND PARTICIPANTS: Consecutive patients presenting to neurology services with new-onset epilepsy or established epilepsy of unknown etiology were identified. Serum samples were tested for autoimmune encephalitis Abs as well as thyroperoxidase (TPO) and glutamic acid decarboxylase 65 (GAD65) Abs. An antibody prevalence in epilepsy (APE) score based on clinical characteristics was assigned prospectively. Data were collected from June 1, 2015, to June 1, 2016.

MAIN OUTCOMES AND MEASURES

Presence of neurological Abs. A score based on clinical characteristics was assigned to estimate the probability of seropositivity prior to antibody test results. Good seizure outcome was estimated on the basis of significant reduction of seizure frequency at the first follow-up or seizure freedom.

RESULTS

Of the 127 patients (68 males and 59 females) enrolled in the study, 15 were subsequently excluded after identification of an alternative diagnosis. Serum Abs suggesting a potential autoimmune etiology were detected in 39 (34.8%) cases. More than 1 Ab was detected in 7 patients (6.3%): 3 (2.7%) had TPO-Ab and voltage-gated potassium channel complex (VGKCc) Ab, 2 (1.8%) had GAD65-Ab and VGKCc-Ab, 1 had TPO-Ab and GAD65-Ab, and 1 had anti-Hu Ab and GAD65-Ab. Thirty-two patients (28.6%) had a single Ab marker. Among 112 patients included in the study, 15 (13.4%) had TPO-Ab, 14 (12.5%) had GAD65-Ab, 12 (10.7%) had VGKCc (4 of whom were positive for leucine-rich glioma-inactivated protein 1 [LGI1] Ab), and 4 (3.6%) had N-methyl-D-aspartate receptor (NMDAR) Ab. Even after excluding TPO-Ab and low-titer GAD65-Ab, Abs strongly suggesting an autoimmune cause of epilepsy were seen in 23 patients (20.5%). Certain clinical features, such as autonomic dysfunction, neuropsychiatric changes, viral prodrome, faciobrachial dystonic spells or facial dyskinesias, and mesial temporal sclerosis abnormality on magnetic resonance imaging, correlated with seropositivity. The APE score was a useful tool in predicting positive serologic findings. Patients who were Ab positive were more likely to have good seizure outcome than were patients with epilepsy of unknown etiology (15 of 23 [65.2%] vs 24 of 89 [27.0%]; odds ratio, 4.8; 95% CI, 1.8-12.9; P = .002). In patients who were seropositive, reduction in seizure frequency was associated with use of immunomodulatory therapy.

CONCLUSIONS AND RELEVANCE

Among adult patients with epilepsy of unknown etiology, a significant minority had detectable serum Abs suggesting an autoimmune etiology. Certain clinical features (encoded in the APE score) could be used to identify patients with the highest probability of harboring neurological Abs.