Diagnostic, prognostic and monitoring lipid biomarkers of multiple sclerosis in human biofluids

Abstract

Molecular biomarkers for multiple sclerosis (MS) are lacking, making early diagnosis, prognosis, and monitoring challenging. Myelin is composed of 70-80% lipid in dry weight, making it an important target for biomarker development in demyelinating conditions. This thesis aims to identify diagnostic lipid biomarkers in cerebrospinal fluid (CSF) and serum that reproducibly differentiate MS from other neuroinflammatory or demyelinating conditions (such as neuromyelitis optica spectrum disorder (NMOSD)) via mass spectrometry, predictive models and machine learning. This work also explores lipid associations with clinical outcomes (relapse status, MRI lesions, cytokine profiles, expanded disability status scale (EDSS), and AI-derived brain volumetrics) to identify biomarkers for monitoring inflammatory activity and demyelination.

We compared, for the first time, CSF lipidomic profiles of relapsing-remitting MS (RRMS) and NMOSD, revealing differences in glycosphingolipids, which yield high diagnostic accuracies in differentiating NMOSD from RRMS. Upon validation, these biomarkers may assist in diagnosing seronegative NMOSD and MS cases. In serum, 20 lipids were reproducibly increased in RRMS compared to controls across two independent cohorts. Associations of acylcarnitine 16:0 and eicosapentaenoic acid-containing lysophospholipids to brain volume changes and EDSS suggest altered energy metabolism, immune activation and inflammation. Upon further validation, these are promising circulating proxies for disability progression and neurological decline that can complement clinical assessment and MRI. Finally, examination of the plasma lipidome after intermittent calorie restriction revealed changes in sphingolipid, lysophospholipid and phosphatidylinositol levels, the latter of which were associated with regulatory T cells. Together, this suggests enhanced lipolysis, anti-inflammatory effects and support for brain lipid homeostasis.