Application of the APE2-CHN and RITE2-CHN scores for autoimmune seizures and epilepsy in Chinese patients: A retrospective study

•The most prevalent neural-specific antibodies detected in our sampling of Chinse patients with seizures or epilepsy of unknown etiology were NMDAR, GABABR, and MOG.•An APE2-CHN score ≥ 5 had 85.5 % sensitivity and 58.9 % specificity in predicting autoantibody positivity.•A RITE2-CHN score ≥ 8 had 98.6 % sensitivity and 83.8 % specificity in predicting positive outcomes in seizure control. To assess the performance of the Antibody Prevalence in Chinese Patients with Epilepsy and Encephalopathy (APE2-CHN) and Response to Immunotherapy in Chinese Patients with Epilepsy and Encephalopathy (RITE2-CHN) scores in Chinese patients with epilepsy of unknown etiology. We conducted a retrospective study of selected patients from Xiangya Hospital, Central South University (01/01/2017-02/28/2019) whose serum and/or cerebrospinal fluid (CSF) samples were examined for autoimmune encephalitis antibodies. Of these, patients with diagnostic code of seizure or epilepsy were selected in our study. An APE2-CHN score was assigned to each patient and a RITE2-CHN score was calculated for each patient who received immunotherapy. Receiver-operating characteristic (ROC) curve was used to assess each score. 191 patients were enrolled in our study. 36 were identified with specific etiologies. The remaining 155 patients had unknown etiology. Central nervous system-specific antibodies were detected in 76 (49.0 %) patients, after excluding patients with only anti-thyroid peroxidase or glutamic acid decarboxylase antibodies. N-methyl-d-aspartate receptor (NMDAR) antibody (48.7 %, 37/76) was the most common subtype of our sample, followed by γ-aminobutyric acid type B receptor (GABAR) (14.5 %, 11/76). Clinical features including new-onset epilepsy, neuropsychiatric changes, speech disorder, movement disorder and inflammatory CSF profile correlated with positive antibody results. The sensitivity and specificity of APE2-CHN ≥ 5 in predicting the presence of neural-specific autoantibodies in our study were 85.5 % and 58.9 % respectively. In the subset of patients who received immunotherapy (n = 112), sensitivity and specificity of a RITE2 -CHN ≥ 8 in predicting favorable seizure outcome were 98.6 % and 63.2 % respectively. The area under the curve (AUC) of RITE2-CHN was greater than that of RITE2 (Z = 3.196, p < 0.05) while there was no significant difference in AUC between APE2 and APE2-CHN (Z = 1.058, p = 0.290). The APE2-CHN and RITE2-CHN scores may be useful screening tools in predicting