A new study set out to identify pathways and upstream regulators that correlate with all measurable aspects of multiple sclerosis types, and to develop and validate cerebrospinal fluid biomarker-based models predicting disability and severity outcomes. The researchers found that compartmentalized inflammation and its effector mechanisms showed the strongest link to MS severity.

Current MS treatments target the immune system and lose their efficacy with the advancing patients’ age of therapy initiation. This has led to broad belief that neuroinflammation decreases with MS evolution and instead neurodegenerative mechanisms cause continuous central nervous system tissue destruction in progressive MS. Such late-disease mechanisms are inadequately reflected in acute and short-term animal models and up to this point were only investigated in postmortem human pathology studies.

Researchers at the National Institute of Allergy and Infectious Diseases, Frederick National Laboratory for Cancer Research, and Novartis Institutes for Biomedical Research analyzed nine prospectively acquired clinical and imaging outcomes in 176 relapsing-remitting and 215 progressive MS patients and 45 healthy volunteers, along with matched cellular and more than 5,000 protein data in 1,042 CSF samples. CSF proteins significantly correlated with the MS outcomes were evaluated by Ingenuity Pathway Analysis to uncover MS-related biology. Machine learning algorithms were employed to generate and independently validate CSF protein-based models predicting nine MS outcomes.

The researchers found that compartmentalized inflammation and its effector mechanisms, such as pyroptosis, showed the strongest link to MS severity, irrespective of clinical categorization of patients. Humoral responses and viral infection-related pathways were linked only to disease severity, not with MS progression. Moreover, localization of the CNS injury showed distinct molecular signatures. While patients with predominant brain injury showed proportionally higher levels of neuroinflammation, spinal cord involvement was reflected by increased fibrosis and tissue hypoxia. CNS-related processes, such as synaptogenesis, were beneficial and no inflammation-unrelated neurodegeneration was identified. CSF biomarker-based ML models predicted nine clinical and imaging outcomes with independently validated performance that exceeds all previously published MS models.

Proteomic data collected from minute volume of CSF reflects diverse molecular processes previously identified only by CNS biopsies or autopsies, and reliably predicts all currently measurable phenotypical aspects of MS in living subjects.

The study was presented at the 2023 ACTRIMS forum.