2026 American Academy of Neurology Abstract Website

David Acero-Garces¹, Moran Guo², Diana Lopez³, Martha Moyano³, Laura Quintero³, Maria Reyes⁴, Jorge Jimenez⁵, Reydmar Lopez⁵, Guillermo Gonzalez-Manrique⁶, Jairo Lizarazo Niño⁷, Federico Silva⁸, Angela Catalina Vallejo Cajigas⁹, Viviana Martinez⁹, Julie Benavides¹⁰, Christian Rojas Cerón³, Juan Rojas³, Gustavo Ramos³, Jonathan Urrego³, Susana Dominguez Penuela¹, Lyda Osorio³, Benjamin Larman¹, Beatriz Parra³, Carlos Pardo-Villamizar¹¹
¹Neurology, Johns Hopkins University, ²Johns Hopkins University, ³Universidad del Valle, ⁴Hospital Simon Bolivar, ⁵Universidad de Antioquia, ⁶Hospital Universitario Hernando Moncaleano Perdomo, ⁷Hospital Universitario Erasmo Meoz, Universidad de Pamplona, ⁸Fundacion Cardiovascular De Colombia, ⁹Hospital Universitario Departamental de Nariño, ¹⁰Universidad Cooperativa de Colombia, ¹¹Johns Hopkins U, Med Dept of Neurology

Objective:

To characterize the antibody reactivity (reactome) against viruses and human peptides, including linear and conformational antigens, in patients with encephalitis, aiming to improve diagnostic precision.

Background:

The humoral response of autoimmune and infectious encephalitis remains poorly characterized. Molecular Indexing of Proteins by Self-Assembly (MIPSA) is a high-throughput assay to generate DNA-barcoded genome-scale libraries of human full-length proteins, human peptides, and viral peptides, allowing comprehensive antibody profiling (reactome analysis).

Design/Methods:

MIPSA was used to identify antibodies to viral (n=285,669) and human peptides (n=353,034) and human full-length proteins (n=15,516) in paired cerebrospinal fluid (CSF) and serum samples from patients diagnosed with encephalitis in the Neuroinfections Emerging in the Americas Study (NEAS) in Colombia and Johns Hopkins Hospital in the USA. These patients were classified as infection-associated, autoimmune, or unknown-cause encephalitis. Cross-validated penalized logistic models were used to predict the etiologic diagnosis by assessing reactivity to signature peptides, based on the reactivity profile of cases with known etiology.

Results:

Fifty CSF-serum pairs from patients with encephalitis were included: 21 infectious (IE), 14 autoimmune (AIE), and 15 of unknown cause. A regularized logistic regression model based on the antibody reactivity against viral peptides classified encephalitis etiology with 88.6% accuracy (CI 73.3%-96.8%) in CSF and 85.7% (CI 69.7-95.2%) in serum. Classification tree models (CART) yielded similar performance. Combining the CSF and blood models, 13 cases of unknown cause were assigned either an infectious (n=9) or autoimmune (n=4) profile. The full-length human protein library yielded lower classification accuracy, at 68.6% (CI 50.7%-83.2%).

Conclusions:

Antibody reactome analysis of linear and conformational antigens using MIPSA enables accurate classification of IE and AIE cases based on their antibody reactivity profiles. Integrating the viral and human antibody reactivity profiles may facilitate reaching an etiologic diagnosis in encephalitis of unknown cause.