Evidence-based diagnostic prediction score for pediatric NMDA receptor encephalitis.

Abstract

Objective: Early diagnosis and treatment of anti-N-methyl-D-aspartate receptor encephalitis (NMDARE) are crucial for a favorable prognosis. Detecting the causative autoantibodies can be challenging. Probable diagnostic criteria are useful in adults less so in children. We aimed to develop a novel diagnostic score for pediatric NMDARE using cohort data.

Methods: We retrospectively analyzed pediatric participants (0-18 years) with suspected autoimmune encephalitis who underwent cerebrospinal fluid analysis for antineuronal antibodies (Abs) between January 2015 and March 2023. Clinical data, including symptoms and laboratory findings, were analyzed. Symptoms were selected through univariate analysis and then analyzed with multivariate logistic regression model. Resulting odds ratios were used to calculate scores. Scoring systems were developed and evaluated with five-fold validation and univariate logistic regression. One scoring system was selected to create a diagnostic prediction score for pediatric NMDARE.

Results: Of the 504 patients, 264 met the inclusion criteria, and 39 tested positive for NMDAR Abs. Comparing clinical symptoms between cohorts and identified 15 variables significantly different (p < 0.05) to create a pediatric NMDARE prediction score. This score showed 82.1 % sensitivity and 82.2 % specificity, with an 8-point cutoff. The area under the curve was 0.888 (95 % confidence interval: 0.838-0.939). A five-fold cross-validation showed a sensitivity of 95.6 %, specificity of 71.4 %, and kappa coefficient of 0.670.

Conclusion: We developed a novel evidence-based diagnostic prediction score for pediatric NMDARE that incorporates specific clinical features and laboratory findings. This score may improve diagnostic accuracy and guide early therapy in children with suspected autoimmune encephalitis.