To evaluate the validity of the 2016 clinical diagnostic criteria proposed for probable anti-NMDA receptor (NMDAR) encephalitis in children, we tested the criteria in a Japanese pediatric cohort.
We retrospectively reviewed clinical information of patients with neurologic symptoms whose CSF was analyzed for NMDAR antibodies (NMDAR-Abs) in our laboratory from January 1, 2015, to March 31, 2019.
Overall, 137 cases were included. Of the 41 cases diagnosed as probable anti-NMDAR encephalitis (criteria-positive) according to the 2016 criteria, 13 were positive and 28 were negative for anti–NMDAR-Abs. Of the 96 criteria-negative cases, 3 were positive and 93 were negative for anti–NMDAR-Abs. The sensitivity of the criteria was 81.2%, specificity was 76.9%, positive predictive value (PPV) was 31.7%, and negative predictive value was 96.9%. Compared with the true-positive group, the false-positive group contained more male than female patients (male:female, 4:9 in the true-positive vs 19:9 in the false-positive group, p = 0.0425). The majority of the cases with false-positive diagnoses were associated with neurologic autoimmunity.
The clinical diagnostic criteria are reliable for deciding to start immunomodulatory therapy in the criteria-positive cases. Low PPV may be caused by a lower prevalence of NMDAR encephalitis or lower level of suspicion for encephalitis in the pediatric population. Physicians should therefore continue differential diagnosis, focusing especially on other forms of encephalitis.
Classification of Evidence
This study provides Class IV evidence that the proposed diagnostic criteria for anti-NMDAR encephalitis in children has a sensitivity of 81.2% and a specificity of 76.9%.