Electroencephalographic biomarkers of antibody-mediated autoimmune encephalitis

To identify electroencephalographic (EEG) biomarkers for different subtypes of antibody-mediated autoimmune encephalitis (AE) and assess their significance in disease severity, treatment response, and prognosis.

The clinical and EEG data from 60 AE patients were analyzed. The relationship between EEG severity in the acute phase and disease severity, treatment response, and prognosis was examined to identify factors contributing to poor outcomes.

The cohort included 60 patients with the following subtypes of encephalitis: anti-LGI1 (22), anti-NMDAR (12), anti-GABA_BR (7), anti-GAD65 (6), anti-MOG (7), anti-Caspr2 (4), and GFAP-A (2). EEG abnormalities were detected in 96.7% of patients, higher than imaging abnormalities (66.7%, p < 0.05). Common EEG features included focal (86.7%) or diffuse (13.3%) slow waves, interictal epileptiform discharges (IEDs) in temporal (46.7%) or extratemporal (15%) regions, and clinical or subclinical seizures (36.7%). During the recovery phase, 92.6% of 27 patients showed significant improvement in EEG patterns, with reduced slow waves and IEDs. Specific EEG patterns were associated with different antibody subtypes. Anti-LGI1 encephalitis had two clinical-electroencephalographic patterns: one was MTLE-like seizure with ictal activity originating from the temporal region; the other was FBDS with ictal EEG showing generalized electro-decremental activity before or at the onset of seizure with extensive infra-slow activity superimposed with EMG artifacts. Anti-NMDAR encephalitis was marked by abnormal background activity, including extreme delta brush, frontotemporal delta activity, diffuse or focal slow waves, with scattered and unfixed IEDs. MOG antibody cortical encephalitis usually presented as diffuse or focal slow waves in unilateral or bilateral hemisphere accompanied by ipsilateral IEDs, sometimes with periodic lateralized epileptiform discharges (PLEDs). Anti-GABA_BR and anti-GAD65 encephalitis usually exhibited slow waves, IEDs and ictal activity involving the temporal regions. The EEG severity grading correlated positively with disease severity (r = 0.547, p < 0.0001) and prognosis score (r = 0.521, p < 0.0001). Further ROC curve and binary logistics regression analysis showed moderate to severe abnormal EEG was a risk factor for poor prognosis (OR = 11.942, p < 0.05), with an AUC of 0.756.

EEG is a sensitive and valuable tool for AE and exhibit common and specific features across different AE subtypes. The severity of EEG abnormalities is a strong predictor of disease outcome.