Utility of the APE 2 Score as a Diagnostic Tool for Autoimmune Encephalitis

ABSTRACT Objective To retrospectively evaluate the diagnostic performance of the Antibody Prevalence in Epilepsy and Encephalopathy (APE 2 ) score relative to clinician‐adjudicated autoimmune encephalitis (AE) and the Graus criteria in a tertiary neuroimmunology referral cohort, including antibody‐negative AE. Methods We conducted a retrospective single‐center study of consecutive referrals to a tertiary neuroimmunology clinic (January 2017–May 2023). AE diagnosis was determined by expert consensus based on clinical features, investigations, exclusion of alternative diagnoses, and immunotherapy response. APE 2 scores and Graus categories, were assigned retrospectively; APE2SM [APE2 seizure/movement] was examined as a post hoc exploratory secondary analysis. Diagnostic performance was assessed using ROC analysis and contingency tables, and logistic regression modeled the probability of AE across scores within this study cohort. Results Eighty‐five patients were included (56.5% female; mean age 53.8 years). Fifty‐nine (69.4%) had AE (34 antibody‐positive). APE 2 demonstrated higher discriminative performance than Graus categories in this cohort (AUC 0.894 vs. 0.804) and in antibody‐negative patients (AUC 0.905 vs. 0.850). In both models, an APE 2 cutoff ≥ 4 balanced sensitivity and specificity near or above 80%, whereas in the full patient cohort no Graus criteria threshold achieved a similar balance. Probability modeling showed APE 2 ≥ 4 crossed 80% in all patients and 60% in antibody‐negative patients within this cohort. Exploratory analysis showed that APE2SM had overall discrimination comparable to the original APE 2 score, with a slightly lower AUC in the full cohort (0.888 vs. 0.894). Interpretation APE 2 score demonstrated good diagnostic discrimination and may represent a useful adjunctive tool for early AE evaluation, including antibody‐negative cases. The exploratory APE2SM model may further enhance sensitivity; however, prospective multicenter validation is needed before routine clinical implementation.