Abstract
Granulocytes play a well-established role in the pathogenesis of brain tissue damage in neuromyelitis optica spectrum disorder (NMOSD). The release of granulocyte activation markers (GAM) into CSF has recently been shown to distinguish NMOSD from multiple sclerosis (MS) with high accuracy. However, their pathogenetic role in myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD) is less clear, and their usefulness for diagnostic differentiation is unknown. This observational cohort study by eight tertiary centres in Europe and Japan included 244 CSF samples from patients with MOGAD (n = 71), NMOSD (n = 48), MS (n = 125) and control persons (n = 19). CSF levels of GAM [neutrophil elastase, myeloperoxidase, neutrophil gelatinase-associated lipocalin (NGAL), matrix metalloproteinase-8 and 9 (MMP-8, MMP-9)], astrocyte damage markers [ADM: glial fibrillary acidic protein (GFAP), S100B], and complement factors C5 and C5a were analysed by capillary ELISA (Ella™) or Luminex®. The primary outcome was the capacity of these markers to differentiate MOGAD, NMOSD and MS in the acute (≤21 days post-exacerbation) stage, and the correlation of GAM with C5 and C5a. Secondary analyses included the correlations of these markers with disability severity, measured by the Expanded Disability Status Scale (EDSS). GAM (except for MMP-9), ADM and C5/C5a levels peaked at onset of disease exacerbation of MOGAD and NMOSD (regardless of aquaporin-4 antibody status), and were significantly higher than in MS. MMP-9 levels were continuously increased in MS over MOGAD and NMOSD, both in acute and subacute/chronic stages. C5 and C5a were equally increased over MS in acute stages of MOGAD and NMOSD. A logistic model and receiver operating characteristics analyses incorporating GAM and C5 displayed high discriminatory power between MOGAD/NMOSD versus MS [area under the curve (AUC) = 0.880], NMOSD versus MS (AUC = 0.837) and MOGAD versus MS (AUC = 0.925) in acute stages. Accordingly, increased ADM levels in NMOSD differentiated NMOSD from MS and MOGAD (AUC = 0.897 and 0.843, respectively). GAM levels correlated with EDSS scores in MOGAD and NMOSD, but not in MS, while those of ADM correlated with disability in NMOSD, but not in MOGAD and MS. Determining CSF levels of GAM and C5/C5a, and of ADM provide a biology-driven approach to differentiate MOGAD, NMOSD and MS. Their measurement can be processed faster and with similar accuracy than with most autoantibody assays, enabling timely initiation of appropriate therapy in acute presentations. The correlation between GAM and C5/C5a levels with neurological impairment in MOGAD and NMOSD corroborates their role as effectors of neural damage, supporting the acute stage use of inhibitors of C5 activation.