Performance of assessment tools in predicting neural autoantibody positivity in patients with seizures.

BACKGROUND: The identification of patients with seizures of unknown etiology who would benefit from neural antibody testing necessitates effective assessment tools. The study aimed to compare the performance of the Antibody Prevalence in Epilepsy and Encephalopathy (APE2) score and the "Obvious" Indications for Neural Antibody Testing in Epilepsy or Seizures (ONES) checklist. We also intended to evaluate whether the performance of the tools varied by types of antibody.

METHODS: Patients diagnosed with epilepsy, seizures, or status epilepticus of unknown etiology at West China Hospital from January 2019 to December 2021 were included. Paired serum/cerebrospinal fluid samples were analyzed for antineuronal and antiglial antibodies. The APE2 score and ONES checklist were applied, and their outcomes were compared to laboratory antibody test results. Possible false positive neuronal antibody results were excluded in sensitivity/specificity analysis reasonably.

RESULTS: A total of 113 antibody-positive and 159 antibody-negative patients were enrolled in sensitivity/specificity analysis. The ONES checklist showed superior sensitivity than APE2 score (95.6 % vs.79.6 %, P < 0.001). Specificity was not statistically different (60.4 % vs. 57.9 %, P = 0.557). The negative predictive value (NPV) of ONES checklist was higher than that of APE2 score (94.8 % vs 80.7 %, P < 0.001). The positive predictive value of them was not statistically different (61.7 % vs 58.8 %, P = 0.557). APE2 score exhibited lower sensitivity for predicting LGI-Abs (52.9 % vs. 80.3 %, P = 0.022) compared to NMDAR-Abs. Similarly, ONES checklist showed lower sensitivity for LGI1-Abs than NMDAR-Abs (82.4 % vs. 100.0 %, P = 0.009).

CONCLUSIONS: The ONES checklist demonstrates superior sensitivity for neural antibody positivity than APE2 score. Specificity of the two assessment tools was similar. ONES checklist performed better NPV than the APE2 score. Both assessment tools performed less well in predicting the presence of LGI1- Abs when compared to NMDAR-Abs.