Serum SERPINA3 as a Candidate Non-Invasive Biomarker for Neuromyelitis Optica Spectrum Disorder: Proteomic Discovery and Same-Center Validation

Background: Diagnostic evaluation of suspected neuromyelitis optica spectrum disorder (NMOSD) integrates AQP4-IgG testing, clinical assessment, neuroimaging, and exclusion of alternative inflammatory demyelinating disorders; nevertheless, some patients remain diagnostically unresolved or may be misclassified as multiple sclerosis (MS) or other disorders. Methods: This single-center retrospective exploratory study analyzed quantitative serum proteomics from 20 patients with MS and 20 with NMOSD. Proteins meeting nominal p < 0.05 together with prespecified fold-change criteria were considered exploratory candidates. VWF, PPBP, and SERPINA3 were subsequently assessed by ELISA in a separate, non-overlapping cohort from the same center, together with routine hematological variables. Results: Among 261 protein entries, 34 met the exploratory nominal threshold and fold-change criteria, but none remained significant after Benjamini–Hochberg correction (lowest q = 0.0734). VWF and PPBP showed no significant differences across the four validation groups. SERPINA3 differed across groups (Kruskal–Wallis H = 44.982, p = 9.34 × 10−10) and was higher in NMOSD than in MS, TBI, and HCs after Holm adjustment (adjusted p = 0.036, 1.49 × 10−9, and 1.13 × 10−5, respectively). In the full ELISA cohort, the AUC was 0.928 for NMOSD versus HC and 0.800 for NMOSD versus MS. In a post hoc sensitivity analysis excluding participants with preceding infection, the NMOSD–MS separation was attenuated. In smaller complete-case analyses, adding an additional laboratory variable to SERPINA3 did not significantly improve apparent discrimination. Conclusions: SERPINA3 is therefore an exploratory adjunctive serum biomarker candidate rather than a stand-alone diagnostic test; prospective external validation in clinically representative cohorts is required.