CSF Study Reveals Distinct Cytokine and Complement Signatures in NMOSD, MOGAD, and MS

Researchers at Johns Hopkins University have mapped the cerebrospinal fluid fingerprints of three immune-mediated central nervous system diseases, finding that the antibody-associated conditions AQP4+ neuromyelitis optica spectrum disorder (NMOSD) and MOG antibody-associated disease (MOGAD) share a distinctive rise in the inflammatory messenger IL-6 that is absent in multiple sclerosis (MS). The study, published in Annals of Clinical and Translational Neurology, also documented widespread elevation of complement cascade components in the spinal fluid of both NMOSD and MOGAD patients during attacks, a pattern that was far less pronounced in MS.

The investigation drew on samples from a single referral center biorepository. The cerebrospinal fluid (CSF) cohort comprised 19 participants with AQP4+ NMOSD, 29 with MOGAD, 36 with MS, and 51 non-inflammatory controls undergoing lumbar puncture for conditions such as idiopathic intracranial hypertension, migraine, and functional neurological disorder. A separate plasma cohort included 18 AQP4+ NMOSD patients, 35 MOGAD patients, 58 MS patients, and 29 healthy controls. All participants with neuroinflammatory disease had shown inflammatory activity within six months before sample collection, defined as a symptomatic attack for NMOSD and MOGAD and as an attack or gadolinium-enhancing lesions for MS. Diagnoses were confirmed against current criteria, including the 2015 International Panel criteria for AQP4-IgG seropositive NMOSD, the 2023 International MOGAD panel criteria, and the 2024 revised McDonald criteria for MS.

The research team measured eleven cytokines, including IFN-gamma, TNF-alpha, IL-6, IL-10, IL-17a, and the B-cell survival factor BAFF, using a multiplexed automated immunoassay. They also quantified the neuroglial proteins neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP), and used sandwich ELISA to measure a broad panel of complement components in CSF, ranging from free C1q and C1s through C2, C3, C4 and their activation products to C5a and the membrane attack complex C5b-9. Group comparisons relied on linear regression models adjusted for age, with false discovery rate correction applied to the primary comparisons against controls; other analyses were exploratory and uncorrected, a caveat the authors emphasize.

The cytokine results revealed a clear split between the antibody-associated diseases and MS. CSF IL-10 and TNF-alpha were elevated in all three disease groups compared with controls but did not differ among the diseases. CSF IL-6, by contrast, was raised in AQP4+ NMOSD and MOGAD relative to both controls and MS, with no difference between the two antibody-associated conditions. CSF BAFF, a cytokine that promotes B-cell survival and activation, was higher in MOGAD than in controls, with a similar but statistically non-significant trend in NMOSD, and was actually lower in MS patients than in controls. Notably, plasma cytokine levels showed no meaningful differences between healthy controls and any disease group, underscoring that the informative signals were confined to the central nervous system compartment.

The biomarker proteins told their own story. CSF and plasma NfL, a marker of neuroaxonal injury, were elevated in all three diseases compared with their respective controls but did not distinguish between them. GFAP, which reflects astrocytic injury, stood out as the signature of NMOSD: it was significantly higher in AQP4+ NMOSD than in every other group in both CSF and plasma, consistent with the known astrocytopathy driven by AQP4 antibodies. GFAP was not elevated in MOGAD, where astrocytes are largely spared, although plasma GFAP was modestly raised in MS compared with healthy controls. Among the 29 participants with paired CSF and plasma samples collected within two days, CSF and plasma levels correlated strongly for NfL, moderately for GFAP and BAFF, and not at all for IL-6, IL-10, IL-17a, or TNF-alpha.

The complement findings were among the most striking. In AQP4+ NMOSD, multiple CSF complement components, including C1s-C1 inhibitor complex, C2, C2b, C3a, C3d, C4a, and C5a, were higher than in controls. In MOGAD, free C1q, C1s, C1 inhibitor, C1s-C1inh, C3, C3a, C3d, C4, and C5a were all elevated compared with controls. The two antibody-associated diseases differed little from each other, except that C4 was higher in MOGAD while C1s-C1inh was higher in NMOSD. In MS, only specific activation products, namely C2b, C3d, C4a, and C5a, were elevated compared with controls, and several components were higher in both NMOSD and MOGAD than in MS. Interestingly, CSF C5b-9, the terminal membrane attack complex, was lower in NMOSD and MOGAD than in controls, which the authors suggest may reflect consumption of complement through activation and deposition within central nervous system lesions rather than free diffusion into the CSF, or the rapid degradation of this labile product.

Correlation analyses linked these immune measures to tissue injury markers and disability. In NMOSD, CSF GFAP correlated positively with IL-6, IL-10, TNF-alpha, and BAFF, and with the complement components free C1q, C1s-C1inh, and C4a, while NfL correlated with TNF-alpha. In MOGAD, both NfL and GFAP correlated with numerous complement components, including free C1q, C1 inhibitor, C1s-C1inh, C2, C2b, C3d, C4, and C4a. In MS, NfL correlated with BAFF and C4a. Clinically, CSF BAFF, GFAP, free C1q, C1s, C1s-C1inh, and C4a correlated with EDSS at nadir in NMOSD, while in MOGAD, IL-10, TNF-alpha, free C1q, C1 inhibitor, and C4a showed similar correlations with nadir disability, and C2 correlated with follow-up EDSS. In MOGAD, IL-6, IL-10, TNF-alpha, BAFF, and several complement components were higher in non-optic neuritis attacks than in optic neuritis presentations. No correlations between CSF analytes and disability emerged in MS.

The authors interpret these patterns as evidence of distinct CSF cytokine profiles in the antibody-mediated demyelinating diseases compared with MS, and of widespread classical complement pathway activation during attacks in both NMOSD and MOGAD. Elevated early classical pathway components, such as C1s-C1inh in NMOSD and free C1q, C1s, and C1 inhibitor in MOGAD, support antibody-driven complement activation, although contributions from other pathways cannot be excluded. The team is careful to note that these data alone cannot establish complement as a pathogenic driver in MOGAD; the elevations may instead reflect secondary consequences of acute inflammation. The role of complement in MOGAD remains uncertain, and prior work has suggested that MOG-IgG may trigger complement-dependent cytotoxicity less efficiently than AQP4-IgG, with histopathological studies showing variable deposition of the membrane attack complex in MOGAD lesions.

The IL-6 findings carry therapeutic resonance. The IL-6 receptor blocker satralizumab is an approved treatment for AQP4+ NMOSD, tocilizumab has shown effectiveness in both NMOSD and MOGAD in observational studies, and a Phase III trial of satralizumab in relapsing MOGAD is ongoing. The authors propose that characterizing complement activation patterns and cytokine profiles may eventually aid differential diagnosis, point to novel therapeutic targets, and illuminate the central nervous system microenvironment in these disorders. They also caution that plasma measurements added little diagnostic value in this study, correlating poorly with CSF levels for most cytokines.

Several limitations temper the conclusions. The sample size was modest, particularly for the rare AQP4+ NMOSD group, and all participants came from a single referral center. Samples were collected up to six months after attack onset, a window in which these dynamic molecules likely fluctuate considerably, although a sensitivity analysis restricted to samples collected within 60 days yielded largely similar results. Some patients had received corticosteroids before sampling, and treatment effects cannot be fully excluded. Plasma samples were often collected later after attack onset than CSF samples, and only about a third of the cohort had paired specimens. The study captured a single time point per patient, some cytokines fell below the detection threshold in many samples, and residual confounding from demographic and clinical differences between the diseases cannot be ruled out. Validation in larger, independent, and ideally longitudinal cohorts will be needed to determine whether these CSF signatures can move from the research bench into clinical practice.

Subject of Research: Cerebrospinal fluid complement and cytokine profiles in AQP4+ NMOSD, MOGAD, and multiple sclerosis during attacks

Article Title: Investigation of Complement Activation and Cytokine Signatures in AQP4+ NMOSD, MOGAD, and MS

Article References: Filippatou, A. G., Dominguez‐Penuela, S., Acero‐Garces, D., Gakis, G., Smith, M. D., Resto, Y., Reyes‐Mantilla, M., Chen, H., Hernandez, I., Said, Y., Salazar‐Camelo, A., Huynh, C., Tassoni, A., Yednock, T., Artis, R., Andrews‐Zwilling, Y., Newsome, S. D., Saidha, S., Bhargava, P., … Sotirchos, E. S. (2026). Investigation of Complement Activation and Cytokine Signatures in AQP4 + NMOSD , MOGAD , and MS. Annals of Clinical and Translational Neurology, Article acn3.70549. https://doi.org/10.1002/acn3.70549

Image Credits: AI Generated

DOI: 10.1002/acn3.70549

Keywords: AQP4+ NMOSD, MOGAD, multiple sclerosis, complement cascade, cytokines, IL-6, BAFF, GFAP, neurofilament light chain, cerebrospinal fluid biomarkers, autoimmune neurology, Johns Hopkins study