2025 American Academy of Neurology Abstract Website

Grace Gombolay¹, James Brenton², Jennifer Yang³, Coral Stredny⁴, Ryan Kammeyer⁵, Kristen Fisher⁶, Alexander Sandweiss⁶, Timothy Erickson⁷, Varun Kannan⁸, Catherine Otten⁹, NgocHanh Vu¹⁰, Jonathan Santoro¹¹, Karla Robles Lopez¹², Robert Goodrich¹³, Scott Otallah¹⁴, Janetta Arellano¹⁵, Andrew Christiana¹⁶, Morgan Morris¹⁷, Mark Gorman¹⁸, Alexandra Kornbluh¹⁹, Ilana Kahn²⁰, Leigh Sepeta²¹, Yike Jiang²², Eyal Muscal²³, Kristy Murray²⁴, Manikum Moodley²⁵, Duriel Hardy²⁶, Claude Steriade¹⁶
¹Emory University/Children'S Healthcare of Atlanta, ²University of Virginia Health System, ³Rady Childrens Hospital/UCSD, ⁴Children's Hospital Boston, ⁵Childrens Hospital Colorado, ⁶Baylor College of Medicine, ⁷Texas A&M University, ⁸Emory/CHOA, ⁹Seattle Children's, ¹⁰Vanderbilt University, Child Neurology, ¹¹Department of Neurology, Children's Hospital Los Angeles, ¹²UNIVERSITY OF TEXAS AT AUSTIN/DMS, ¹³Wake Forest Baptist Medical Center, ¹⁴Wake Forest, ¹⁵CHOC, ¹⁶NYU, ¹⁷Emory University, ¹⁸Boston Children's Hospital, ¹⁹Children's National Hospital, ²⁰Childrens National Medical Center, ²¹Children'S National Health System, ²²Duke University, ²³Baylor, ²⁴Emory University School of Medicine, ²⁵Dell Children's, ²⁶Dell Children's Specialty Pavillian

Objective:

To assess whether electroencephalography (EEG) features associate with one-year outcomes in anti-NMDA receptor encephalitis (NMDARE).

Background:

Electroencephalography (EEG) is abnormal in ~90% of patients with anti-NMDA receptor encephalitis.^1-3 However, limited data is available on EEG characteristics associated with poor outcomes in pediatric NMDARE (pNMDARE).

Design/Methods:

We performed a multi-site retrospective cohort study (CONNECT study) from 14 institutions in the United States, including children under 18 years of age diagnosed with pNMDARE.^4,5 The modified Rankin Score (mRS) was used to define good (mRS ≤ 2) and poor (mRS ≥ 3) outcomes at one year from onset. Initial EEG features obtained from EEG reports were included. Univariate and multivariate logistic regressions were performed with R CRAN (v4.4.1).

Results:

In this cohort (n=249), 226 (91%) children had initial EEG data and one-year mRS available. An abnormal EEG was noted in 200/246 (81%), with findings that included: background slowing (154/246=63%), epileptiform discharges (81/246=33%), and extreme delta brush (13/217=6%). Seizures were recorded in 39/246 (16%) and status epilepticus in 18/246 (7%). The only EEG feature associated with one-year outcomes on univariate analysis was background slowing (p=0.043). On multivariate analysis, background slowing on EEG had an odds ratio of 5.5 (95% 2.4–12.9) were associated with poor one-year outcomes when adjusting for ICU admission and non-improvement within four weeks of immunotherapy initiation. The proportion of abnormal EEG in routine versus prolonged EEGs was not significant (p=0.056), but seizures and status epilepticus were more likely to be detected on prolonged EEGs (p=0.0003 and p=0.0009, respectively) as compared to routine EEGs, which could either reflect that these features are detected on a longer EEG versus a longer EEG was ordered due to the seizures and/or status epilepticus.

Conclusions:

Alterations in the EEG background was associated with poor one-year outcomes, even after adjusting for ICU admission and improvement within four weeks.